WIP: Refactor folding ranges to use emitter pattern (alternative approach)

[andyw8]

No description provided.

[github-actions]

Benchmark results in seconds (slowest at top)

          textDocument/completion average: 0.265179 std_dev: 0.005053
          textDocument/diagnostic average: 0.039816 std_dev: 0.008755
          textDocument/definition average: 0.004348 std_dev: 0.002877
      textDocument/selectionRange average: 0.003378 std_dev: 0.000461
            textDocument/codeLens average: 0.003143 std_dev: 0.000111
        textDocument/documentLink average: 0.00313 std_dev: 0.000119
 textDocument/semanticTokens/full average: 0.003097 std_dev: 0.000357
      textDocument/documentSymbol average: 0.003091 std_dev: 0.000122
   textDocument/documentHighlight average: 0.001804 std_dev: 0.000121
textDocument/semanticTokens/range average: 0.000843 std_dev: 4.7e-05
               codeAction/resolve average: 0.000635 std_dev: 5.8e-05
               textDocument/hover average: 0.000596 std_dev: 5.7e-05
           textDocument/inlayHint average: 0.000593 std_dev: 2.6e-05
          textDocument/formatting average: 6.7e-05 std_dev: 0.000301
    textDocument/onTypeFormatting average: 6.6e-05 std_dev: 4.2e-05
          textDocument/codeAction average: 3.8e-05 std_dev: 2.8e-05
        textDocument/foldingRange average: 3.3e-05 std_dev: 7.0e-06


================================================================================
Comparison with main branch:

 textDocument/semanticTokens/full slower by 87.343 %
textDocument/semanticTokens/range unchanged
      textDocument/documentSymbol slower by 84.913 %
        textDocument/foldingRange faster by 97.907 %
          textDocument/formatting unchanged
          textDocument/diagnostic unchanged
        textDocument/documentLink slower by 81.986 %
           textDocument/inlayHint unchanged
      textDocument/selectionRange unchanged
   textDocument/documentHighlight unchanged
               textDocument/hover unchanged
          textDocument/codeAction unchanged
    textDocument/onTypeFormatting unchanged
               codeAction/resolve unchanged
          textDocument/completion unchanged
            textDocument/codeLens slower by 81.905 %
          textDocument/definition unchanged


At least one benchmark is slower than the main branch.


================================================================================
Missing benchmarks:

RubyLsp::Requests::ShowSyntaxTree

[andyw8]

Superseded by #1013