Add csvmark component for CSV ↔ Markdown table conversion

[Copilot]

New Wasm component that converts between CSV documents and GitHub-flavored Markdown tables, with correct escaping for embedded commas, quotes, pipes, newlines, and UTF-8.

WIT

package yoshuawuyts:csvmark;

world csvmark {
    export csv-to-md: func(input: string) -> result<string, string>;
    export md-to-csv: func(input: string) -> result<string, string>;
}

Implementation (components/csvmark/src/lib.rs)

• csv crate parses input; first record becomes the markdown header, ragged rows are padded to the widest row.
• pulldown-cmark (with ENABLE_TABLES) walks events to extract the first table; Soft/HardBreak inside cells collapse to spaces.
• Cell escaping mirrors tablemark: | → \|, line breaks → space; md-to-csv reverses the pipe escape and lets csv::Writer handle CSV-side quoting.

Tests

11 unit tests in lib.rs covering simple conversion, embedded commas + escaped quotes, pipe/newline escaping, UTF-8 (Cyrillic), empty input, ragged-row padding, CSV quoting of special fields, pipe unescaping, and a CSV → MD → CSV round trip.

Wiring

• justfile: added cargo build -p csvmark to build; csvmark added to publish and versions recipes.
• .github/workflows/publish.yml: csvmark added to the target choice list, included in the publish-component job guard, and given a description.
• README.md: row added to the components table.

Notes

• Crate layout, edition = "2018", and empty tests/test.rs intentionally mirror tablemark/wordmark for consistency.
• crate-type is ["cdylib", "rlib"] (vs. tablemark's cdylib-only) so unit tests can link against the lib on the host target; the wasm artifact is unaffected.