gh-118761: improve optparse import time by delaying textwrap import

[eli-schwartz]

The same change was made, and for the same reason, by argparse back in
2017. The textwrap module is only used when printing help text, so most invocations will never need it imported.

See: #1269
See: 8110837

• Issue: Improve import time of various stdlib modules #118761

[AA-Turner]

Style: move local imports to the top of the block

Suggested change

	text_width = max(self.width - self.current_indent, 11)
	indent = " "*self.current_indent
	import textwrap
	import textwrap
	text_width = max(self.width - self.current_indent, 11)
	indent = " "*self.current_indent

[eli-schwartz]

Sure, makes sense!

For what it's worth the reason I did it this way was a direct consequence of having just been reading argparse.py, which breaks this style suggestion, while documenting the motivation of the change. :) I suppose it is not worth bothering with the churn there as it's just a style nit on code that isn't being modified already...

[AA-Turner]

Style: move local imports to the top of the block

Suggested change

	help_text = self.expand_default(option)
	import textwrap
	import textwrap
	help_text = self.expand_default(option)

[picnixz]

Out of curiosity, how are the hyperfine and -X importtime benchmarks?

[eli-schwartz]

As mentioned in the other PR -- I'm not used to benchmarking and I may be doing something totally wrong because I can get results saying this PR is now slower to import, or results saying it's faster, and hyperfine warns that I have lots of "outliers". But here's a favorable result.

Produced by copying the old and new versions of the file so that I can import them both without switching branches:

$ LD_LIBRARY_PATH=$PWD hyperfine --warmup 8 "./python -c 'import optparse_old'" "./python -c 'import optparse_new'"
Benchmark 1: ./python -c 'import optparse_old'
  Time (mean ± σ):      23.6 ms ±   7.7 ms    [User: 18.4 ms, System: 5.1 ms]
  Range (min … max):    17.2 ms …  47.7 ms    92 runs
 
  Warning: Statistical outliers were detected. Consider re-running this benchmark on a quiet system without any interferences from other programs. It might help to use the '--warmup' or '--prepare' options.
 
Benchmark 2: ./python -c 'import optparse_new'
  Time (mean ± σ):      17.5 ms ±   2.8 ms    [User: 14.4 ms, System: 3.0 ms]
  Range (min … max):    16.3 ms …  31.1 ms    98 runs
 
  Warning: The first benchmarking run for this command was significantly slower than the rest (29.2 ms). This could be caused by (filesystem) caches that were not filled until after the first run. You are already using the '--warmup' option which helps to fill these caches before the actual benchmark. You can either try to increase the warmup count further or re-run this benchmark on a quiet system in case it was a random outlier. Alternatively, consider using the '--prepare' option to clear the caches before each timing run.
 
Summary
  ./python -c 'import optparse_new' ran
    1.35 ± 0.49 times faster than ./python -c 'import optparse_old'