[phase-31 1/4] Sort schema foundation

[g-talbot]

Summary

Sort schema types, parsing, and time-window arithmetic for metrics compaction (Phase 31 Metadata Foundation, PR 1 of 4).

Stacks on matthew.kim/metrics-wide-schema.

What's included

quickwit-proto:

• Vendored event_store_sortschema.proto (SortSchema, SortColumn, RowKeys, ColumnValue types)
• SortFieldsError error type for parser/validation errors
• From<SortFieldsError> impl for MetastoreError

quickwit-parquet-engine:

• sort_fields/ module:
  • column_type.rs — ColumnTypeId enum matching Go iota values, suffix/name resolution
  • parser.rs — Direct port of Go StringToSchema with V2-only enforcement
  • display.rs — schema_to_string / schema_to_string_short (port of Go SchemaToString)
  • equivalence.rs — Schema comparison for compaction (port of Go EquivalentSchemas)
  • validation.rs — Schema validation rules (port of Go ValidateSchema)
  • window.rs — window_start() with rem_euclid for negative timestamp correctness, validate_window_duration() with hour-divisibility constraint
  • tests.rs — 97 tests including Go test port, proptests for window invariants
• table_config.rs — TableConfig with per-product-type default sort fields, serde support

Verification

• cargo build -p quickwit-parquet-engine ✅
• cargo test -p quickwit-parquet-engine -- sort_fields table_config ✅ (97 tests)
• cargo clippy -p quickwit-parquet-engine --all-features --tests ✅ (no new warnings)

Test plan

• All 97 sort_fields and table_config tests pass
• Proptest properties verified (window alignment, containment, determinism)
• Round-trip tests for parse → display → parse
• Go test suite fully ported
• Clippy clean

🤖 Generated with Claude Code

[fulmicoton-dd]

@mattmkim I only reviewed the changes in the global scope. Can you complete the review?