feat: add counters to bulk rename function

[JustForFun88]

This Pull Request introduces counter-based bulk renaming to Yazi, enabling users to specify incrementing placeholders across multiple lines. Each line can include zero or more placeholders of the form:

%{<COUNT_TYPE>,<START_VALUE>,<STEP>,<WIDTH>}

where:

• <COUNT_TYPE>: A single character indicating the counter type:
  • N, n, D, d → Numeric digits.
  • A → Uppercase ANSI letters.
  • a → Lowercase ANSI letters.
  • R → Uppercase Roman numerals.
  • r → Lowercase Roman numerals.
  • C → Uppercase Cyrillic letters.
  • c → Lowercase Cyrillic letters.
• <START_VALUE> (optional): Initial value, either as a number (e.g., 1, 2, etc.) or a value corresponding to the counter type:
  • A, B, AA (ANSI uppercase)
  • a, b, aa (ANSI lowercase)
  • I, II, III (Roman uppercase)
  • i, ii, iii (Roman lowercase)
  • А, Б, АБ (Cyrillic uppercase)
  • а, б, аб (Cyrillic lowercase)
  • _ for unspecified.
• <COUNT_STEP> (optional): Step size, integer (e.g., 2) or _ for unspecified.
• <COUNT_WIDTH> (optional): Minimum width with zero-padding. Integer (e.g., 3) or _ for unspecified.

Optional parameters (<START_VALUE>, <STEP>, <WIDTH>) can be set or left as _ to take defaults. For instance, %{n,_,2,3} starts at the default value 1, increments by 2, and has a minimum width of 3. They also can be omitted, i.e. not written at all. For example, you can write %{N}, %{N, 1}

Optional parameters must be specified sequentially: <START_VALUE> is required if <COUNT_STEP> or <COUNT_WIDTH> are used, either explicitly (e.g., 1) or with _. Omitting earlier parameters with commas (e.g., %{N,,2} or %{N,,,4}) is invalid. User should write %{N,_,2} or %{N,_,_,4} instead.

Escaping: To insert a literal %{ in filenames (in linux for example), user could escape it by writing %%{. For example, file_%%{name}.txt will become file_%{name}.txt.

Constraints & Error Handling:

• If any line contains counters, all lines must have the same number of placeholders; mismatches result in a comprehensive error report showing all problematic lines simultaneously.
• All errors are collected and reported together.

To avoid adding extra dependencies, the parser is custom-written. While parsing performance could theoretically be improved using Rayon parallelization, I think that typical usage scenarios likely won’t require such optimization.

default.mp4