fix(library): read footprint and lib-table fields from the parsed tree - #322

Summary

get_footprint_info and parse_lib_table read their string fields by scanning
the raw file text instead of the S-expression they had already parsed. Two
user-visible consequences:

• An escaped quote truncates the value. KiCad writes a description of
  0.1" pitch headers as (descr "0.1\" pitch headers"); the scan stopped at
  the escaped quote and returned 0.1\.
• A field written across a line break is missed entirely, and the entry comes
  back blank — an fp-lib-table entry loses its nickname, URI and type, so its
  libraries stop resolving.

Not linked to an issue. #84 tracks the same root cause elsewhere but explicitly
excludes parse_lib_table, whose indentation scan was already fixed; this is
the field scraping that remained inside each located block.

Approach

extract_sexp_string looked for the literal (key " and read to the next ":

let pat = format!("({} \"", key);
let start = block.find(&pat)? + pat.len();
let end = block[start..].find('"')? + start;

It has no notion of escapes, of nesting, or of any whitespace other than one
space. Both call sites now read from the parsed tree — find_str(tag) for a
tagged field, and the root node's datum for a footprint's name — so escape
decoding and whitespace tolerance come from the parser. The function had no
callers left and is deleted.

Two get_footprint_info hunks are behaviour-preserving and included only for
consistency: find_all("pad").len() becomes an explicit direct-child filter
(find_all is already direct-children-only), and !find_all("model").is_empty()
becomes find("model").is_some(), which short-circuits and allocates nothing.

Why lib-table entries are still located textually. The obvious move is
parse_sexp over the whole file. parse_lib_table is infallible by design —
every caller path (read_flat_lib_table → flatten_lib_table) degrades to an
empty list and a tracing::warn! rather than surfacing an error — so a
whole-file parse would let one malformed entry silently discard every library
in the table, which for this path means footprint and symbol resolution failing
wholesale. Keeping find_block_starts + find_balanced_block (both already
string-aware and indentation-agnostic) preserves per-entry tolerance while still
removing the scraping. A malformed entry is now skipped with a warning naming
its offset.

Compatibility and safety

No public compatibility impact. No tool, argument, schema, config key or path
changes. Read paths only — nothing here writes a file or touches IPC.

get_footprint_info's name and description response fields can now return a
correct value where they previously returned a truncated or empty one; no key
is added, removed or renamed.

Validation

$ cargo fmt --all -- --check
$ cargo test --workspace --locked --lib --tests
$ cargo test --workspace --locked --doc
$ cargo clippy --workspace --locked -- -D warnings

All four clean, run against this branch on top of main (not merely on the
development branch this was cherry-picked from).

The two bug-fix tests were verified to fail against the old scraper before the
fix was restored:

parse_lib_table_keeps_an_escaped_quote_in_a_description ... FAILED
parse_lib_table_reads_a_field_split_across_lines ... FAILED

parse_lib_table_skips_only_the_entry_that_does_not_parse passes both before
and after — it is not a regression test for a fixed bug, but a guard on the
tolerance property against the whole-file-parse refactor described above. Its
comment says so.

Not run: viewer, plugin, packaging and real-KiCad checks. This touches one file
in konnect-core and none of those paths.

• cargo fmt --all -- --check
• cargo test --workspace --locked --lib --tests (what CI runs)
• cargo test --workspace --locked --doc
• cargo clippy --workspace --locked -- -D warnings
• Relevant viewer, plugin, packaging, and real-KiCad checks — not applicable

Review checklist

• The diff is focused and contains no generated output, personal data, or unrelated cleanup.
• New names follow docs/NAMING_CONVENTIONS.md; public renames include compatibility handling.
• New behavior and failure paths have regression coverage.
• File mutations are atomic and preserve unrelated content — read paths only, nothing written.
• IPC mutations verify the requested board and do not leave partial batches — no IPC.
• If tools were added/removed: counts and docs updated — no tools added or removed.