added some pure-python performance optimizations

[shakfu]

I've added some pure-python optimizations which improve performance by 30% or so.

1. Pre-compile regex patterns at module level instead of compiling on every function call
2. Use list accumulation with "".join() instead of O(n²) string concatenation
3. Cache dynamically-generated regex patterns to avoid repeated compilation

These are standard Python optimization techniques that require no additional dependencies.

I've also added the docs/optimizations.md document to explain these in more detail.

[jhidding]

Oh, this is so cool! Thank you, also for the extensive documentation and explanations. :)

About the caching of regexes: I read in the documention that Python is already performing this sort of caching on its own, but I never tested this.

[shakfu]

Oh, this is so cool! Thank you, also for the extensive documentation and explanations. :)

My pleasure.

About the caching of regexes: I read in the documention that Python is already performing this sort of caching on its own, but I never tested this.

I didn't know they were cached either!