Mech Vs tank combat in a one on one is determined by one thing, weapons. If you try to take on a tank with something like: Gauss, PPC, ATMs, MRMs, or Large Lasers you will fail, hard.
Those weapons are all excellent at killing mechs but fail horribly at killing tanks. If you have those weapons, even if you outweigh your opponent, you should not engage with confidence.
RACs, Pulse lasers, SRMs, and Heavy lasers on the other hand, wreck tanks. So long as you're not standing still taking a pounding from the tank, you will win.
Its all about DPS against tanks, whereas mechs are more vulnerable to burst and splash damage.
This is both a product of the engine (the splash damage weapons will only hit one location on a tank usually because the locations are massive) and a product of design (everything will still do advertised damage, but some weapons are far more effective).
The larger tanks already are basically support vehicles, great in base defense and attack but poop out in the field.
The exception being the Morrigu, which pays for this with its exceptional price and extreme vulnerability to bad terrain.
Also, pure speed numbers are misleading. A damaged mech can run away and weave through cover while protecting its back if it needs to disengage. The tank can only turn around and run, exposing its rear armor the entire time. You can't weave and you cant try and lose opponents in heavy cover like forests or very rough terrain because you will be the one getting slowed down or stuck not your pursuers.
Expecting a group of tanks to "hold its own" versus ONE heavy mech is never going to happen, at least not with our supported player numbers. In a 32v32 or 64v64? Sure why not. But 14v14?
No way, that is a third of the team.
I will say that I'm not a fan of the 100->0 damage output of tanks, I would rather them lose their offensive effectiveness overtime rather than their mobility overtime due to damage. Their mobility is already piss poor, having to crawl back at 20kph is cancerous.
Take a UAC20 or RAC5 hollander (even perhaps the LBX hollander might work), you will dump on an Ares/chevy easily and maybe even partisan/oro if you're good or lucky, but that HVAC10 sniper? That will be awful against all tanks, solely because of how the weapon works. Same for the Gauss variants, even the Hgauss is poor against tanks in a straight up fight
The Osiris is quite good at early game brawling. The uller sadly, is having some serious problems. The Chevy very much is a mini demo and is meant to be. It has very little for weapons and some of the weirdest handling in the game (its the only tank that is better in the field than in a base). But the thought about the harasser being a mini demo.... come on now man, that's just nonsense. The hull is so low and transfer rates so high that if you lose a single location on a harasser you will get killed by one or two stray shots, it happens that fast. You have to move extremely fast and drive like a maniac to survive in those things.
The solitaire will get some armor and hitbox fixes so that will be better at brawling. I'm doing everything I can to make the uller not die horribly (but this is very difficult). Right now the Owens is in a good place, the MRM10 one does nearly as much damage as the old SRM6 model, just not a point blank. Speed is life for all light assets, I get the feeling you just want to be able to ram into enemies and straight up tank and then run away with lol speed (like the old starter owens) which is just cancer all around. I think people are careful early game because they're not very confident in lights, piloting errors will lead to death very quickly and right now there is more close range firepower on the field early game than long range. You just have to be careful to close in properly and no when to gtfo.
If you want to just run in and loltank... then take the chevy... thats what its there for. The owens can do it too but stacking a bunch of brawl weapons on it gets expensive quickly.
All of these "weird" values are by design and all accomplish very specific purposes.
The chevy's bizarre armor distribution are a product of its bizarre shape and the fact that it cannot turn in place like a traditional tank can. Even trying to present your front or rear to the enemy in a chevy you will most of the time be presenting one of your sides because the its so much longer than it is wide and you need to be constantly moving, which means lots of turning to avoid running into things.
All of what you state is by design, its not weirdness at all.
If the turret had just as high of transfer rate as the sides or rear on the large tanks, it would be completely pointless to go for anything other than the turret; even with those liberal numbers its often STILL the best option to just for for the turret.
Yet on the chevy, the turret is oddly tiny and out of the way, again its actually hard to hit it if you try in a battle, so packing on a bunch of armor to something that never gets hit is a ticket to garbage asset.
The concept of unified design across all assets is possible in many ways, trying to stay to tabletop rules, but there are a few major exceptions.
The biggest exception of all, which I would tell anyone trying to make a new mechwarrior game, is that the literal shape of the asset MUST determine its armor distribution.
The numbers MUST bow to the reality of what the player sees in game, if not, the asset will always be garbage and always fail, regardless of its dice roll reputation.
Heavy armor redistribution has saved the Uziel, Mauler, Partisan, Puma, Cougar, Chimera, Avatar, Hollander, Rifleman, Oro, and Chevalier from being regulated to complete shit.
Hell, even the ATLAS sucked before I decided to tweak the armor and put in a standard fusion engine which has a much lower transfer rate.
I'm probably forgetting some others here too, there is a small elite of logically designed mechs that are very streamlined and hard to focus, they benefit from the standard armor rules, anything that has a distinctive shape in anyway will suffer.
The fact that you all have concerns of things being "a bit too OP" or UP or what have you, is a good sign. Because its very difficult to that point as is, there is a whole lot of backroom magic going on to make sure whatever you take will feel at least 90% worth it if you use it correctly while still not breaking any hard CBT logic. Sure mechlab would still probably break the funness of the game (not sure it would break "balance" perse) but there are LOTS of variants in the game and I push silly boats within reason to make sure everything has drawbacks. The fact that there is no 100% best thing everytime consensus is a good thing.
Even the Lanner, which is currently, quite heavily bugged, is no god.