Apply @r-pkg-optimizer skill to stretch out some performance improvements

[MichaelChirico]

cc @dshkol - https://github.com/dshkol/rpgkoptskill

Optimizer output:

---

This PR optimizes performance and memory usage across lintr's core execution pipeline without modifying public APIs, S3 classes, or linter semantics:

1. Lazy Function Call Caching in build_xml_find_function_calls() (R/source_utils.R):

   • Defers XPath evaluation of //SYMBOL_FUNCTION_CALL/parent::* and //SLOT/parent::expr[...] so it only runs on demand when xml_find_function_calls() is queried, avoiding eager evaluations on AST subtrees that never use function-call indexing.
   • Replaces <<- with an encapsulated environment (cache_env) to adhere to package style guidelines.

2. Pre-partitioned Parse Data Subsetting in get_source_expressions() (R/get_source_expressions.R):

   • Replaces repeated $O(N \times M)$ linear scans (parsed_content[which(top_level_map == id), ]) with a single-pass index grouping via split(seq_len(nrow(parsed_content)), factor(top_level_map, levels = tl_ids)).

3. Package Namespace & S3 Generic Memoization (R/namespace.R):

   • Adds .namespace_cache to memoize results for namespace_imports(), imported_s3_generics(), and exported_s3_generics().
   • Keys cache entries by package path and NAMESPACE file modification timestamp (file.mtime()), eliminating hundreds of redundant file re-reads and namespace scans during multi-file linting while preserving real-time invalidation when files change.

4. Short-Circuit Empty Nodesets in strip_comments_from_subtree() (R/xml_utils.R):

   • Adds if (length(expr) == 0L) return(expr) to avoid executing xml_find_first_(expr, ".//COMMENT") on empty nodesets across 12 linters.

5. Streamlined Linter Orchestration in lint_impl_() (R/lint.R):

   • Pre-partitions active linters into expression-level and file-level subsets once per file to avoid redundant per-expression necessary_linters() overhead, and avoids accumulating 0-length lint lists.

---

Benchmarks & Performance Results

1. Microbenchmarks & Core Pipeline (bench::mark)

Workload	Metric	Baseline	Optimized	Relative Change
get_source_expressions("R/lint.R")	Median Latency	147.0 ms	112.0 ms	-23.8% (-35 ms)
	Memory Allocations	26.6 MB	20.3 MB	-23.7% (-6.3 MB)
	Throughput	6.17 itr/s	7.67 itr/s	+24.3%
	Garbage Collections	31	24	-22.6%
lint("R/lint.R")	Median Latency	1.05 s	0.965 s	-8.1% (-85 ms)
	Memory Allocations	33.1 MB	26.5 MB	-20.0% (-6.6 MB)
	Throughput	0.914 itr/s	0.976 itr/s	+6.8%
	Garbage Collections	200	178	-11.0%
S3 Generics Resolution (152 files)	Wall-Clock Time	5.66 s	1.77 s	-68.7% (3.2x faster)

2. End-to-End Codebase Linting: lint_dir("R") (152 files)

Metric	Baseline	Optimized	Difference
Elapsed Wall-Clock Time	34.55 s	27.74 s	-19.7% (-6.81 s)
Lints Found	1,348	1,348	Exact Match

3. Evaluation on Heavy & Massive R Source Files

Workload	Lines / Size	Expressions	get_source_expressions()	lint() Total Time	Lints Found
QC.R (tools)	10,654 lines / 421 KB	411	1.63 s	13.62 s	5,551
tests.Rraw (data.table)	21,987 lines / 1.26 MB	15,344	16.75 s (1.09 ms/expr)	444.10 s	47,353

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 99.73%. Comparing base (c3c9805) to head (024974a).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@             Coverage Diff             @@
##              main    #3111      +/-   ##
===========================================
- Coverage   100.00%   99.73%   -0.27%     
===========================================
  Files          129      129              
  Lines         7468     7533      +65     
===========================================
+ Hits          7468     7513      +45     
- Misses           0       20      +20     

☔ View full report in Codecov by Harness.
📢 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.