Tighten up docs on hash mismatch

[kasbah]

These are a bit too wordy for a situation where we normally "just want to get the check to pass". Also they talk about "instructions given below the error message" which don't seem to exist.

[jmoggr]

Those instructions appear for hash mismatches that are not related to pnpm. pnpm.fetchDeps rolls their own hash checking, so it has a non-standard error message. I'll open a follow up PR with this clarification.

[kasbah]

Wait, why do we set the pnpmDeps.hash to empty? I just ran it without doing that and it tells me the hash that I need to copy in.

$ nix build .#frontend
error: hash mismatch in fixed-output derivation '/nix/store/4wpp80j18vvm232ii1ajl2kqnbfgvzq2-frontend-pnpm-deps.drv':
         specified: sha256-vuBwhtNTTRbpgPZS+AQDybASYM9rwWYG8l0bscVQUso=
            got:    sha256-sxczRF8IsYqQzmAAv+IiFeWJygqHCVnSk8fEuy5d1JM=

[jmoggr]

Yeah... we only need to do that for non pnpm.fetchDeps hashes, since pnpm.fetchDeps is special (with nicer behavior!). At the time I wrote the docs there were more "standard" hash mismatch issues and I muddled up the issues in the docs. The instructions will still get you the same results.

[jmoggr]

I'm working on the fixes, I'll push them here.

[jmoggr]

I pushed the changes. The pnpm specific docs are improved and we kept the docs for other hash mismatches.

[kasbah]

Somehow, later on a different branch, I was not getting the "got" hash when I was doing this again and had to put in empty string. I'm really not sure what I was doing differently.

[jmoggr]

I vaguely recall seeing something like that at some point, but dismissed it since there were bigger nixy changes. I'll have claude try to grind out a reproduction.

[kasbah]

We could just put the advice back to put in the empty string since that always works even if it's a tiny bit more work.

[epatters]

Feel free to merge when you're satisfied.