Prevent identifiers longer than 64 characters

[zachmu]

Also fixed a bug where we allowed multiple column names with the same case-insensitive name

Fixes dolthub/dolt#6611

[max-hoffman]

This improves correctness but the implementation is counterproductive to the ongoing refactors. Validation should happen before query rewrites. For ex, you don't need a new rule to validate the ident when converting ast.RenameTable to plan.RenameTable. Named rules are useful when series of seemingly OK individual transforms in combination create problems. Validation doesn't benefit from rule names for context, errors are typed and provide all of the necessary context.

[max-hoffman]

changes to this file make PRs in GItHub harder to read and usually cause merge conflicts

[zachmu]

This improves correctness but the implementation is counterproductive to the ongoing refactors. Validation should happen before query rewrites. For ex, you don't need a new rule to validate the ident when converting ast.RenameTable to plan.RenameTable. Named rules are useful when series of seemingly OK individual transforms in combination create problems. Validation doesn't benefit from rule names for context, errors are typed and provide all of the necessary context.

Discussed offline. I initially made this change in part, but on consideration this pattern of erroring on statements during parse that used to parse correctly can be very dangerous for backwards compatibility, since existing databases definitions will no longer parse without error in contexts where they need to. To move more validation logic into the parser safely, we need to also support permissive parsing modes to prevent existing databases from tripping newly added validations.