[3.0] Replace NameTrimmer with equivalent IdentifySharedPrefixes mod

[Exanite]

Summary of the PR

(WIP)

This PR chains off of #2555 and focuses on rewriting the NameTrimmer class as a new IdentifySharedPrefixes mod with equivalent functionality. The goal is to keep all outputs the same while removing edge cases and generally just making the code easier to understand.

Note: Part of the reason this PR chains off of #2555 is because #2555 actually needs some of the changes here to fully work correctly. #2555 can be merged as is, but one unit test has to be commented out and a non-critical feature will be missing.

Related issues, Discord discussions, or proposals

This is the conversation in January when this change was briefly discussed: https://discord.com/channels/521092042781229087/587346162802229298/1457864604951777441

Further Comments

Tasks

• Add a IdentifySharedPrefixes mod that identifies shared prefixes and uses the name affix system to annotate those prefixes.
  • Ideally rewrite to avoid the "3 pass" logic. I have no idea how it works still.
  • Ideally avoid using Humanizer and use the new NameSplitter code for consistency (especially since we have thorough test coverage for this code)
• Remove prefix trimming special casing from PrettifyNames
  • Visitor
  • Prettify-only pipeline
• Rewrite rest of trimming+prettify pipeline as sequential INameProcessors
  • This should include the name conflict resolution code, since that's currently done as a step right after the name processors run.
  • This will act as the main and only code path. Currently we have two code paths (which technically was 3 before I removed another one during [3.0] Improve handling of nested struct names and affixes #2555).
  • Remove the name container isolation and just provide all names together. If name processors need the isolation, we still provide the container names and container members separately and can implement the old behavior as needed, but restricting access to all names is unnecessary and can actually lead to incorrect behavior.
• Change INameProcessor to be an interface private to PrettifyNames
  • We can change this later if required, but I currently don't see a point to exposing it. There are plenty of ways to configure the name transformation process already through ways that are more declarative and configurable by the user.
• Add unit tests documenting all common and known edge cases for IdentifySharedPrefixes
  • This is an initial list. I need to verify this behavior against the old behavior.
  • Simple shared prefix, no configured prefix hint
  • Simple shared prefix, with shorter prefix hint (the hint should be preferred)
  • Case where a bit more than half of the names share a prefix (or whatever the breakpoint is)
  • Multiple shared prefixes, with multiple configured hints
  • Case where a prefix is somewhat shared, but not quite due to capitalization/underscores/etc
    • Pretty sure we should normalize here, then identify accordingly
    • Also ensure that we don't naively use the "shared" prefix in the NameAffix attribute. We use the shared prefix to determine the actual prefix, which is going to be a variation of the shared prefix in this case.
  • Migrate the old test cases over from the NameTrimmer tests
    • This should cover some edge cases such as trimming names into nothing (due to a name being a prefix of another name)
• Regenerate all bindings using Windows and ensure no changes to output