perf: optimize from_str_radix

[DaniPopes]

Use actual from_str routines instead of treating every character like a u64 digit.

[codspeed-hq]

Merging this PR will improve performance by ×32

⚡ 59 improved benchmarks
✅ 327 untouched benchmarks

Performance Changes

	Benchmark	BASE	HEAD	Efficiency
⚡	most_significant_bits/4096/4096	28.9 µs	26.1 µs	+10.88%
⚡	parse/binary/256/max	51.7 ms	15.2 ms	×3.4
⚡	parse/binary/192/max	33.5 ms	11.4 ms	×2.9
⚡	parse/binary/192/zero	598.8 µs	248.7 µs	×2.4
⚡	parse/binary/128/zero	559.6 µs	198.3 µs	×2.8
⚡	parse/binary/4096/max	77.9 ms	2.4 ms	×32
⚡	parse/binary/64/zero	467.5 µs	139.6 µs	×3.3
⚡	parse/binary/384/max	104.1 ms	22.8 ms	×4.6
⚡	parse/binary/128/max	19.6 ms	7.6 ms	×2.6
⚡	parse/binary/256/zero	679.8 µs	287.9 µs	×2.4
⚡	parse/binary/4096/zero	62.5 µs	32.4 µs	+93.11%
⚡	parse/binary/512/zero	1,061.9 µs	511.6 µs	×2.1
⚡	parse/binary/512/max	171.4 ms	30.3 ms	×5.7
⚡	parse/binary/64/max	8.5 ms	3.3 ms	×2.6
⚡	parse/decimal/128/max	6.3 ms	1.6 ms	×3.9
⚡	parse/decimal/192/zero	598.8 µs	273.7 µs	×2.2
⚡	parse/binary/384/zero	875 µs	410.9 µs	×2.1
⚡	parse/decimal/192/max	10.4 ms	2.2 ms	×4.7
⚡	parse/decimal/128/zero	559.6 µs	215.1 µs	×2.6
⚡	parse/decimal/256/max	16.1 ms	3.3 ms	×4.8
...	...	...	...	...

ℹ️ Only the first 20 benchmarks are displayed. Go to the app to view all benchmarks.

---

_{Comparing DaniPopes:dani/from-str-radix-perf (89f64de) with main (c147f59)}