Immutable elementor and parallel MRT records parsing

[ties]

Elementor took &mut self when convering a MrtRecord to BgpElem. This branch has a version that does not require mutating state.

This allows an Elementor to be shared between threads after the peer index table is set.

examples/parallel_records_to_elem.rs shows how this can be used and shows a speedup. In my testing, the speedup is significant. This feels logical to me, the heavy work of parsing the content can now be done on many threads.

$ cargo run --release --example parallel_records_to_elem
    Finished `release` profile [optimized] target(s) in 0.07s
     Running `target/release/examples/parallel_records_to_elem`
Total number of routes: 55539269
Time elapsed: 28.961109239s
Total number of records: 1356921
Time elapsed: 2.206397117ss (to chunk into raw records)
Total records: 55539269
Time elapsed: 3.712137851s

[codecov]

Codecov Report

❌ Patch coverage is 0% with 24 lines in your changes missing coverage. Please review.
✅ Project coverage is 91.09%. Comparing base (5eb9107) to head (fb48926).
⚠️ Report is 13 commits behind head on main.

Files with missing lines	Patch %	Lines
src/parser/iters/mod.rs	0.00%	24 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #262      +/-   ##
==========================================
- Coverage   91.23%   91.09%   -0.15%     
==========================================
  Files          83       83              
  Lines       14939    14963      +24     
==========================================
  Hits        13630    13630              
- Misses       1309     1333      +24     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[Copilot]

Pull request overview

This PR introduces a non-mutating (“immutable”) Elementor conversion path to enable sharing an initialized Elementor across threads and parsing MRT records into BgpElems in parallel, primarily to speed up large RIB/TableDumpV2 processing.

Changes:

• Added Elementor::record_to_elems_iter(&self, ...) plus new lazy iterator types (RecordElemIter, BgpUpdateElemIter) and explicit ElemError for non-mutating conversion.
• Added convenience APIs on BgpkitParser to return an Elementor pre-initialized from an initial PeerIndexTable along with record/raw-record iterators for parallel processing.
• Added an example and changelog entry documenting the parallel parsing approach and expected performance gains.

Reviewed changes

Copilot reviewed 6 out of 6 changed files in this pull request and generated 6 comments.

Show a summary per file

File	Description
src/parser/mrt/mrt_elem.rs	Introduces immutable/lazy element conversion APIs and iterator implementations; refactors existing Vec-based conversion to reuse iterator path.
src/parser/mod.rs	Re-exports new public iterator/error types from mrt_elem.
src/parser/iters/mod.rs	Adds helper methods to pre-extract PeerIndexTable and return (Elementor, iterator) pairs for parallel workflows.
src/parser/iters/default.rs	Tweaks record filtering logic when filters are present.
examples/parallel_records_to_elem.rs	New example demonstrating multi-threaded parsing with shared immutable Elementor.
CHANGELOG.md	Documents the new immutable/lazy Elementor APIs and the parallel processing example.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[Copilot]

RecordElemIter::RibAfi::size_hint returns a lower bound of 0 even though entries is an IntoIter with an exact remaining length. This underestimates iteration size and can lead to missed preallocation opportunities. Consider returning (entries.len(), Some(entries.len())).

[Copilot]

In RecordIterator::next, this branch is only reached when filters is non-empty (due to the earlier if filters.is_empty() { ... } else { ... }). The added self.parser.filters.is_empty() || check is therefore redundant and makes the filter logic harder to follow; consider simplifying to just the any(match_filters) check here.

Suggested change

	if self.parser.filters.is_empty()
	|| elems.iter().any(|e| e.match_filters(&self.parser.filters))
	{
	if elems.iter().any(|e| e.match_filters(&self.parser.filters)) {

[Copilot]

The format string appends a literal s after {:?} for a Duration. Duration's Debug output already includes the unit suffix (e.g., 2.20s), so this will print ...ss. Consider removing the extra s or formatting with as_secs_f64() if you want a numeric seconds value.

Suggested change

	"Time elapsed: {:?}s (parser -> RawRecord -> Elementor::record_to_elems)",
	"Time elapsed: {:?} (parser -> RawRecord -> Elementor::record_to_elems)",

[Copilot]

New public APIs (record_to_elems_iter, bgp_to_elems_iter, bgp_update_to_elems_iter) don’t appear to have direct unit tests, while this module already has tests for record_to_elems. Adding tests that validate iterator output equivalence vs the existing Vec-based methods, plus the new error cases (UnexpectedPeerIndexTable, MissingPeerTable), would help prevent regressions in this parsing-critical code.