Use Sequence lengths when determining validity for connections

[joeyballentine]

[RunDevelopment]

This check is trivial to defeat in 2 ways:

1. By having a regular node in between
   
2. By having the connection be valid at first, and then changing the sequence length of either node to make it invalid.
   

[joeyballentine]

making sequence length types get passed down to things you have connected sounds like a separate effort / separate PR to me

The other part sounds like I just have to do this same kind of check in the other place we're checking validity? But I thought it used the same check for both? I'll have to do some digging

[joeyballentine]

Yeah I'll be honest here -- I just tried to do the second suggestion and I cannot figure it out. I'm confused on how the current input errors even work, and how i could do something similar for sequence mismatches. There doesn't seem to be any input-to-output matching going on and yet somehow it knows the types are incompatible. I don't get it

[joeyballentine]

TBH, both of these things feel like they should be separate efforts. This just adds the validation for connectivity (since there's 0 validation right now), and in separate PRs we can do what's necessary for getting these other things working

[joeyballentine]

Those two concerns are now addressed in #2996 and #2997

[RunDevelopment]

Blocked by #2996.

This PR doesn't work correctly until #2996, so I won't merge this PR until I can test it.