Enhance strformat library with Python-compatible features

[kobi2187]

Add several Python Format Specification Mini-Language features:

• Sign-aware padding (= alignment): places padding after sign/prefix but before digits, e.g. "{-42:0=10}" -> "-000000042"

• Thousands separators (, and _): group digits with separators

  • Comma (,) for decimal: "{1234567:,}" -> "1,234,567"
  • Underscore (_) for all bases: "{255:_b}" -> "1111_1111"
  • Binary/octal/hex use groups of 4, decimal uses groups of 3

• Percentage type (%): multiplies by 100 and adds % suffix "{0.25:.2%}" -> "25.00%"

• Character type (c): converts integer to Unicode character "{65:c}" -> "A"

Zero padding with grouping properly integrates zeros into the grouped digits for both integers and floats, matching Python behavior.

Added comprehensive test suite based on Python's format specification.

[Araq]

Tell your AI model to use more helper procs, less return, optimize string slices to direct character copy loops and in general follow the existing style. And even after all that is done it looks like pure technical debt, nobody asked for "even more python like f-strings". Already this one-char-based mini-language is undecipherable nonsense, Nim allows for real DSLs, Nim is not Python.

[kobi2187]

thought you'd be happy we make the stdlib more complete. There's a lot of python code out there, when one wants to port to Nim, it can save a lot of trouble if it has higher compatibility. and afaik, the string format came from python, but i could be wrong.