Remove single-character property name optimization from SDDM

[jasonkuhrt]

Problem

SDDM currently uses single-character property names (f, nt, a, etc.) as a bundle size optimization. However, analysis shows this provides minimal benefit after gzip compression:

• Uncompressed savings: 466 bytes (3.6%)
• Gzipped savings: 34 bytes (2.3%)

The short property names hurt developer experience during debugging and code exploration.

Proposal

Replace single-character names with readable names:

• f → fields
• nt → namedType
• a → arguments
• ad → argumentsDescendant
• it → inlineType
• k → kind
• n → name
• fcs → fieldsContainingScalars

Impact

• Bundle size: +34 bytes gzipped (negligible)
• DX: Significantly improved debugging and code readability
• Changes needed:
  • Update type definitions in src/docpar/core/sddm/SchemaDrivenDataMap.ts
  • Update generator in src/generator/generators/SchemaDrivenDataMap.ts
  • Update consumers in src/requestPipeline/encode.ts, src/requestPipeline/decode.ts, parser context, etc.

Analysis

Modern gzip compression largely neutralizes the benefit of short property names because it compresses repetition. After the first occurrence of a property name, subsequent occurrences compress to ~1-2 bytes regardless of length.

The optimization is a micro-optimization that provides minimal real-world benefit while hurting developer experience.

[jasonkuhrt]

Completed in PR #1450 which migrated SDDM to descriptive property names and extension pattern.