Added Sinhala test cases for spaCy, implemented si_tokenizer and si_vocab, and fixed version issues in the requirements file.

[AshenWELI]

---

title: Adding Sinhala (si) language test cases

Description

Previously, spaCy had no test coverage for the Sinhala (si) language.
This PR adds a comprehensive test suite for the Sinhala tokenizer and lexical
attributes, covering real-world text sourced from BBC Sinhala news articles.

Changes introduced

New file: spacy/tests/lang/si/test_text.py
spacy/tests/lang/si/test_tokenizer.py
spacy/tests/lang/si/__init__.py
Edited file spacy/tests/conftest.py

The following test cases were added:

• test_si_tokenizer_handles_long_text — validates correct tokenization
  of a long real-world Sinhala news paragraph (83 tokens), sourced from
  BBC Sinhala.

• test_si_tokenizer_handles_cnts — parametrized tests covering a variety
  of Sinhala sentence structures including punctuation, numbers, question marks,
  semicolons, and mixed numeric-Sinhala tokens (e.g. 110,000කට).

• test_lex_attrs_like_number — validates the like_num lexical attribute
  for Sinhala numeric tokens including digits, fractions, comma-formatted numbers,
  Sinhala cardinal number words (e.g. එක, දෙක), and large number words
  (e.g. බිලියනය).

• test_si_lex_attrs_like_number_for_ordinal — validates like_num for
  Sinhala ordinal forms including suffix-attached ordinals (තුන්වන, සියවන)
  and space-separated ordinals (100 වන, 23 වෙනි).

• test_si_lex_attrs_capitals — confirms that like_num handles Sinhala
  words correctly regardless of .upper() calls, since Sinhala script has no
  concept of upper or lower case.

• test_si_tokenizer_handles_exception_cases — validates tokenizer
  exception handling for punctuation splitting, ordinal numbers (10වෙනි
  staying as one token), and conjunct character sequences (ශ්‍රී, ත්‍රස්තවාදීන්).

Notes

• Sinhala has no upper or lower case — .lower() and .upper() calls have no
  effect on Sinhala script, which is reflected in the test design.
• All test sentences were sourced from authentic Sinhala news media to ensure
  real-world coverage.
• The like_num function in lex_attrs.py was extended to support Sinhala
  ordinal suffixes (වන, වෙනි) and additional cardinal number words.

Types of change

New feature - adds missing test coverage for the Sinhala (si) language module.

Checklist

• I confirm that I have the right to submit this contribution under the project's MIT license.
• I ran the tests, and all new and existing tests passed.
• My changes don't require a change to the documentation, or if they do, I've added all required information.