ci(workflow): upgrade actions and compiler checks

[sambacha]

This addresses #532

Action lint check:

https://rhysd.github.io/actionlint/#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

Questions:

Shouldn't foundry test be using a defined seed value?
What about checking different EVM versions across SOLC versions?
Why not perform gas checks and persist gas snapshots as an artifact, so any regressions can be identified?

Simplify

Doing something like this would make adding EVM versions to the testing easier

jobs:
  build:
    runs-on: ubuntu-latest
    strategy:
      matrix:
     # These values match the names of test groups that will be run
        test-group:
          [
            0.8.0,
            0.7.6,
            0.7.0,
            0.6.12,
            0.6.2,
          ]
      fail-fast: false
    steps:
	# (...)
	# (...)
      - name: Check compatibility with ${{ matrix.test-group }}
        id: compat
        if: always()
        run: |
          output=$(forge build --skip test --use solc:/${{ matrix.test-group }}/)
          if echo "${output}" | grep -q "Warning"; then
              echo "${output}"  >> "$GITHUB_OUTPUT"
              exit 1
          fi

      - name: Pass or fail
        run: |
          if [[ ${{ steps.compat.outputs.output }} == 0 ]]; then exit 0; else exit 1; fi

[mds1]

Presumably this change is to prevent the job running on forks?

[mds1]

Shouldn't foundry test be using a defined seed value?
What about checking different EVM versions across SOLC versions?
Why not perform gas checks and persist gas snapshots as an artifact, so any regressions can be identified?

These are all good questions. If we do address them let's do it in a separate PR, but TLDR is I think the current CI setup is sufficient coverage:

1. Getting different random values in CI each time is ok here to get more coverage so a seed seems unnecessary
2. The number of combinations here can get pretty big with the last few EVM versions, and the solidity team is good about keeping things backwards compatible (i.e. if our code compiles with cancun it will compile with shanghai and paris), so just checking the default one is ok
3. We aren't gas sensitive in tests so this will mostly just add noise/friction to the contribution process IMO

[mds1]

Why do we need this?

[mds1]

Superceded by #550 which is a bit smaller diff