What market cap restrictions do MMHF’s have?

The short answer is depends on GMV book size. Nice rule of thumb to think about is that you start to eat slippage costs once you step beyond 3-5% ADV in a single name - of course not an end all be all rule but helpful to frame up how pod PMs will think about it. Maybe you could have a hard cutoff set at launch by the risk team that the PM agrees to but this is typically a judgement call made by the PM from what I've seen and heard. Don't know about SM's and you mention Citadel so will keep focused on pod structure. This has been my experience so someone from some other pod & PM may have a totally different way of thinking about it. 

Example using round numbers (not perfect):

• If the PM runs a $1B GMV book and the name you're talking only trades $25M a day on average it's going to take 8 days for every 1% of GMV allocation ($1B * 1% = $10M) / ($25M * 5% = $1.25M) - or 4 days if it trades $50M if you want to assume that. So it'll take you 8 days to get in (on 5% ADV) to build a position but more importantly it'll take you 8 days to get out if something happens and your thesis is now broken. Most PMs will not want to sit there watching their traders trying to get out for >trading week while the name moves against them (or you can decide to go over 5% ADV and you're going to be moving the name down yourself off of flows). 
• Another thing to consider is what're they going to have to short against this name - are there good hedges of similar factor exposure or is the PM going to have to short much larger cap names against it since there isn't another similar name of size and now the book is off balance on the size factor and have to go short an index like IWM. Hedging with indices has it's drawbacks and is sometimes the only option but not the most ideal set up. This is really simplified but not going to go down the rabbit hole on factor exposure, etc.  
• Last major thing to consider is opportunity cost - there is a threshold level of work & time to reach the critical mass required to be ready to put the name in the book. If I can only make this a 1% position because of name-specific liquidity why wouldn't I pass on this and look at another idea that's much larger and therefore can be a larger allocation if it turns out being something we really like (all else equal)  

This is talking about capacity from an industry perspective but has similar concepts.

https://twitter.com/FundamentEdge/status/1730624608578269518