feat(analyzer): Support delimited and ASR comma-grouped Turkish National IDs (TCKN) in TrNationalIdRecognizer #2451

Is your feature request related to a problem? Please describe.
TrNationalIdRecognizer (TR_NATIONAL_ID) currently defines a replacement_pairs parameter in __init__ and invokes EntityRecognizer.sanitize_value(pattern_text, self.replacement_pairs) in validate_result, indicating an intent to support formatted/delimited Turkish National ID numbers (TCKN). However, there are two issues in the current implementation:

1. Delimited and ASR comma-grouped TCKNs are missed (replacement_pairs is dead code): self.replacement_pairs defaults to [], and PATTERNS only contains r"\b[1-9][0-9]{10}\b" (11 contiguous digits). As a result, TCKNs formatted with spaces (100 000 001 46), hyphens (100-000-001-46), or comma-separated digit groups produced by Speech-to-Text / ASR models during speaker pauses (100, 000, 001, 46) never match the regex and never reach validate_result.
2. Dummy repeated-digit sequences pass the checksum: Mathematical placeholder sequences where the first 10 digits are identical and the 11th digit is 0 (e.g., 11111111110, 22222222220, ..., 99999999990) satisfy the NVI modulo-10 equations ((5d * 7 - 4d) % 10 == d and 10d % 10 == 0) and are currently promoted to 1.0 confidence because invalidate_result is not implemented.

Describe the solution you'd like
Enhance TrNationalIdRecognizer in presidio_analyzer/predefined_recognizers/country_specific/turkey/tr_national_id_recognizer.py:

• Set default replacement_pairs to [("-", ""), (" ", ""), (",", "")] when replacement_pairs is None (while still respecting an explicit replacement_pairs=[]).
• Add a delimited pattern supporting Turkish TCKN digit groupings (3-3-3-2, 3-2-2-2-2, and 4-4-3) separated by spaces, hyphens, or ASR pause commas (", " / ","), bounded with negative lookbehind/lookahead assertions so partial slices of longer delimited number sequences are not matched.
• Implement invalidate_result (and validation checks) to reject inconsistent/mixed delimiters (e.g., 100-000 001-46) and dummy repeated-digit sequences (11111111110–99999999990).
• Add unit and YAML configuration-path tests (RecognizerRegistryProvider + AnalyzerEngine) in tests/test_tr_national_id_recognizer.py.

Describe alternatives you've considered

• Relying on custom user patterns per deployment, which leaves the built-in replacement_pairs parameter in TrNationalIdRecognizer non-functional out of the box and fails to filter dummy repeated-digit TCKNs.

Additional context
In Turkish call-center ASR (e.g. Whisper) transcripts and formatted documents, 11-digit TCKNs are commonly grouped as 3-3-3-2 (100 000 001 46, 100-000-001-46, or 100, 000, 001, 46). Supporting these groupings while enforcing delimiter consistency and boundary checks prevents both false negatives on speech transcripts and false positives on unrelated numeric sequences.