some refactoring of io formatter internals

[ManuLinares]

extracted logic for hex-buffers (%h), string-padding (%s), hex-floats (%a), and collection printing into helper functions.

---

I did some testing:

Function	Before	After	Change
std::io::out_str	1747	1227	-29.8%
std::io::vprintf	1441	1146	-20.5%
std::io::floatformat	2012	1603	-20.3%
std::io::floatformat_hex	0	496	New
Total	5200	4472	-14.0%

Opt Level	Before (ms)	After (ms)	Difference
O0	63.9 ± 1.6	62.8 ± 1.6	-1.7%
O1	19.2 ± 0.7	18.1 ± 0.7	-5.7%
O2	17.6 ± 0.7	18.2 ± 0.8	+3.4%
O3	17.4 ± 0.6	17.1 ± 0.8	-1.7%
O4	17.2 ± 0.7	17.7 ± 0.7	+2.9%
O5	18.5 ± 0.7	18.2 ± 0.7	-1.6%
Os	17.9 ± 0.7	18.7 ± 0.7	+4.5%
Oz	17.9 ± 0.8	17.8 ± 0.6	-0.6%

• O0/O1: Noticeably faster
• O2/O4/Os: Slightly slower, maybe due to inlining decisions or function call overhead trade-offs.
• O3/O5/Oz: Roughly the same or slightly faster.

I know I probably shouldn't be over-optimizing but oh well

[lerno]

Thank you!