Migrate Spoom to use Rubydex

Spoom performs its own indexing to power a few different features from metrics collection to dead code detection.

1. Code metrics (CodeMetricsVisitor): collects a bunch of metrics about the code and gradual typing usage:

   • number of classes, modules, singleton classes (with and without type members)
   • number of RBS scope annotations (@abstract, @interface, etc.)
   • number of methods (singleton or not)
   • number of attribute accessors
   • number of signatures (sig or RBS comment)
   • number of T. calls
   • number of RBS inline assertion comments

   This can all be replace by querying the global graph if we support RBS comments.

2. Model representation (Model): builds a complete representation of the declarations, definitions and their relationship form a codebase. This is the meat behind dead code detection and any "smart" analysis that needs advanced features like constant resolution or inheritance linearization.

   I believe this whole part of the codebase should be replace 1:1 by Saturn. This is going to be an invasive change.

3. Dead code detection (Deadcode) this comes with many sub-features:

   • Indexing both .rb and .erb files
   • Indexing declarations by unqualified names
   • Indexing references to unqualified names (both constants and methods)
   • Plugin system to listen to events such as entering a definition or entering a call

There is some work in progress here: https://github.com/Shopify/spoom/blob/at-test-indexer/lib/spoom/deadcode/index.rb and early results are really encouraging with a 5x speed increase and 2x less memory.

Start by reading lib/spoom/sorbet/metrics/code_metrics_visitor.rb, lib/spoom/model/model.rb, and lib/spoom/deadcode/index.rb. Compare the existing dead-code index with the at-test-indexer branch and identify how Rubydex or Saturn would cover metrics, model representation, and dead-code indexing. Done means these indexing responsibilities are migrated without losing the listed features.