perf(split): optimize FixedWidthNumber Display implementation

[CrazyRoka]

Currently, the Display implementation for FixedWidthNumber collects all mapped digits into a temporary String before writing them to the formatter. This incurs an unnecessary heap allocation every time a number is formatted.

This change replaces formatting logic with direct write_char calls with optimization in mind.

[CrazyRoka]

There is 0.56% performance improvement. I see room for more optimizations here.
Apart from that, we might need more in depth benchmark, because existing one reports more than 50% wall time in locale initialization and clap building

[codspeed-hq]

Merging this PR will improve performance by 3.98%

⚡ 1 improved benchmark
✅ 283 untouched benchmarks
⏩ 38 skipped benchmarks¹

Performance Changes

	Mode	Benchmark	BASE	HEAD	Efficiency
⚡	Simulation	cp_large_file[16]	381.3 µs	366.7 µs	+3.98%

---

_{Comparing CrazyRoka:optimize-split-number-format (58c32e8) with main (da7ad84)}

Footnotes

1. 38 benchmarks were skipped, so the baseline results were used instead. If they were deleted from the codebase, click here and archive them to remove them from the performance reports. ↩