perf(loro-internal): remove quadratic slow paths in text import/checkout (#893)

* perf: opt step 1

* perf(loro-internal): speed up InsertText slicing during change split

- Add ASCII/suffix fast paths in InnerListOp::slice to avoid O(n^2)
  unicode->byte scans
- Add ignored perf test reproducing split-suffix slicing pattern

* perf(loro-internal): batch id_to_cursor updates on richtext leaf splits

- Batch leaf-split remaps via IdToCursor::update_insert_batch to reduce
  per-split InsertSet::update churn.
- Add InsertSet::update_many fast-path for small fragments (<=
  MAX_FRAGMENT_LEN) by rebuilding runs from a dense map.
- Ignore .trace_analysis/ artifacts.

* fix: a failed fuzzing test

* perf(loro-internal): optimize fast path for plain-text deltas in richtext state

- Introduced a new method `content_node_len` to retrieve the length of content nodes.
- Enhanced the logic for determining when to apply fast updates for plain-text deltas, reducing unnecessary rebuilds.
- Improved performance by minimizing BTree mutations during incremental updates.

* chore: changeset

Author: zxch3n

.changeset/big-suns-share.md

@@ -0,0 +1,6 @@
+---
+"loro-crdt": patch
+"loro-crdt-map": patch
+---
+
+perf(loro-internal): remove quadratic slow paths in text import/checkout #895

.gitignore

@@ -9,6 +9,7 @@ node_modules/
 .idea/
 coverage/
 trace-*.json
+.trace_analysis/
 .loro
 loom_test.json
 .env