fix: transform name capturing regex once

[JLHwung]

Q	A
Fixed Issues?	Fixes #10393
Patch: Bug Fix?	Yes
Major: Breaking Change?
Minor: New Feature?
Tests Added + Pass?	Yes
Documentation PR Link
Any Dependency Changes?
License	MIT

When runtime && !isRegExpTest(path) is true, a new RegExp node will be inserted to AST and visited later by the transform-named-capturing-groups-regex again. Most of cases when regex consists of only named capturing groups, the transformed regex node will be prevented by indexOf("(?<") early return, however when there is a lookbehind <?=, babylon will recurse infinitely.

Here we revise indexOf("(?<") early return by looking ahead(?< to see if it is a lookbehind pattern.

[babel-bot]

Build successful! You can test your changes in the REPL here: https://babeljs.io/repl/build/11462/

[nicolo-ribaudo]

What about using .search(/\?<(?!=/)/) instead of indexOf, to avoid unnecessarily running the regexp transformer?

[JLHwung]

@nicolo-ribaudo Good idea, I revise your suggested regex to include both positive and negative lookbehind.