feat: add support for preprocessing sources

[grandizzy]

Implements a preprocessor allowing us to skip recompiling tests on non-interface source file changes. See the companion Foundry PR foundry-rs/foundry#10010 for benchmarks and more details.

Supersedes #198:

• uses solar AST instead solang-parser to compute interface representation
• uses solar HIR instead of solc AST to find bytecode dependencies
• handles more edge cases, like deploying with salt/value

Closes #197.

TODO

• use solar HIR (deps to review and include feat(hir): expose hir from parsing context paradigmxyz/solar#210)
• unit tests
• opt-in caching changes
• autodetect mocks
• figure out how to parse when building from different dir
• additional testing (tested with uniswap v4-core and simpler projects)

### Depends on

• merge solar PR (feat(hir): expose hir from parsing context paradigmxyz/solar#210)
• solar release

[grandizzy]

the win panic is related to the fact in compilers we normalize sources when reading but then we read them again from disk in Solar locs get out of sync. e.g. with this commit e16fe89 the preprocessor test that panic before pass (https://github.com/foundry-rs/compilers/actions/runs/14091283691/job/39468341793#step:7:708) however there are some flatten related test failing now
@DaniPopes could you pls suggest a proper fix for this, should we change the parsing logic and pass the sources we already read? CC @klkvr

thanks both!

[DaniPopes]

You can manually load sources with SourceMap::new_dummy_source_file and add them with ParsingContext::add_file

"dummy" isn't right, it's just a way to intern a source file with a String rather than loading from disk

edit: removed "dummy" here 4eefc93

[grandizzy]

You can manually load sources with SourceMap::new_dummy_source_file and add them with ParsingContext::add_file

"dummy" isn't right, it's just a way to intern a source file with a String rather than loading from disk

Awesome, that worked, committed in 446a5d8