Basic memory use

[masklinn]

Just informative, mostly to make sure that the regex implementation is not too problematic given ua-parser/uap-rust@a6c3036

This is a slice of time -l running python -mua_parser bench samples/useragents.txt

	run time (seconds)	maximum rss	peak memory footprint
3.12-basic	33.61	42041344	30148544
3.12-basic-sieve-5000	11.67	45400064	29673280
3.12-basic-sieve-50000	9.67	55623680	30000960
3.12-re2	2.74	81657856	29476864
3.12-regex	1.32	163594240	29214528
graalpy-basic	107.61	1542012928	29181824
graalpy-basic-sieve-50000	37.27	1110573056	30590976
pypy-basic	230.88	119734272	29624384
pypy-basic-sieve-50000	42.30	130744320	30296064

so...

1. I have no idea what the hell macos's "peak memory footprint" means (it's lower than the maximum rss, and basically the same for all implementations which makes no sense)
2. the regex implementation is in fact quite memory intensive, this is consistent with previous findings (ua-parser/uap-rust@1af15c1) surprisingly running more stuff on top does not actually fix the issue in any way
3. although caches are not free, and they don't come close to the speed of a native implementation, they turn out not to have that much of an overhead in the grand scheme of things¹²

Footnotes

1. although the bench currently pessimises regex quite a bit, and re2 a bit (to a lesser extent) because it uses eager matchers rather than lazy, so it compiles every regular expression in the set even though regex needs none of that, maybe I should add a toggle for that ↩

2. also wtf graal why do you use less memory with a cache? ↩