Streamline validation

[grarco]

Describe your changes

Closes #3190.
Closes #3193.

This PR fixes some wrong documentation for the finalize_block and check_proposal_tx_function.

It also introduces a new local configuration parameter to allows a node to rerun the process_proposal checks before finalize block in case they don't want to blindly trust the majority of validators. This is implemented for both full nodes and validators since the latter could also skip the process_proposal call in some cases.

Indicate on which release or other PRs this topic is based on

#3192 (diffs for review: https://github.com/anoma/namada/pull/3448/files/176754fbd523ec7d17a1dbbd10f59900b846aeb1..1a704b679b122656aa4ff6d0baa947054c326271)

Checklist before merging to draft

• I have added a changelog
• Git history is in acceptable state

[codecov]

Codecov Report

Attention: Patch coverage is 30.48780% with 114 lines in your changes missing coverage. Please review.

Project coverage is 53.87%. Comparing base (879a326) to head (fc3241d).
Report is 4 commits behind head on main.

Files	Patch %	Lines
crates/node/src/lib.rs	0.00%	49 Missing ⚠️
crates/apps_lib/src/cli.rs	0.00%	28 Missing ⚠️
crates/node/src/shell/mod.rs	31.70%	28 Missing ⚠️
crates/node/src/shell/process_proposal.rs	46.66%	8 Missing ⚠️
crates/apps_lib/src/config/mod.rs	0.00%	1 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3448      +/-   ##
==========================================
- Coverage   53.92%   53.87%   -0.05%     
==========================================
  Files         317      317              
  Lines      107575   107664      +89     
==========================================
- Hits        58011    58007       -4     
- Misses      49564    49657      +93     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[grarco]

NOTE: currently, the recheck is done calling process_txs but I believe it would be slightly better to call process_proposal (which is the handler of the cometbft's call), so that if we changed anything in the handler that would immediately translate to the recheck logic too, while, at the moment, if the modification lies outside of process_txs this would not translate to the recheck. On top of that, because of the shim, nodes can currently call process_txs twice for the same block proposal. With the addition of the recheck logic this can go up to 3 calls to this function: would be nice if we could keep track of the whether the function already ran so that we can ensure that it gets called at most once.

I'll try to address both of these in another branch

[tzemanovic]

1 small style recommendation - other than that LGTM