Allow ebcdic 2.x

[adamzr]

• Issue #: Allow ebcdic 2.x #478
• Have you listed any changes to install or build dependencies? Yes — requirements.txt: ebcdic>=1.1.1,<2 → ebcdic>=1.1.1,<3.
• Ensured your changes are compatible with Python 2.7 (ONLY FOR v0.29). N/A (this PR targets next-release).
• Have you updated the CHANGELOG with details of changes to the codebase, this includes new functionality, deprecated features, or any other material changes.
• If necessary, have you bumped the version number? We will usually do this for you. (Left for maintainers.)
• Have you included py.test tests with your pull request. (Not yet necessary) — no new tests; existing tests.py suite passes locally with ebcdic==2.0.1 installed.
• Ensured your code is as close to PEP 8 compliant as possible? flake8 . --select=E9,F63,F7,F82 clean.
• Ensured your pull request is to the next-release branch (or v0.29 if applicable)?

Summary

Resolves #478. Per the ebcdic changelog, the 2.0 release is "a pure technical release that does not change the functionality of the package" — it only modernizes the build toolchain and drops support for Python 2 and 3.8.

Because extract_msg still supports Python 3.8 (per setup.py and the CI matrix in .github/workflows/python-package.yml), this PR widens the upper bound instead of requiring ebcdic>=2. The lower bound is unchanged, so on Python 3.8 pip will continue to resolve to ebcdic 1.1.1; on Python 3.9+ it can now also pick up 2.x.

This is analogous to #469 / #472 for beautifulsoup4.

Verification

• extract_msg/encoding/__init__.py uses ebcdic only as import ebcdic as _ to register codecs on import. Confirmed all EBCDIC code pages referenced in _CODE_PAGES (e.g. cp1140, cp1141, cp1148ms, cp290, cp420, cp424, cp833, cp838, cp870, cp1097, cp1149, cp273, cp277, cp278, cp280, cp284, cp285, cp297, cp1025) resolve via codecs.lookup() with ebcdic==2.0.1 installed.
• python tests.py: 20 passed, 2 skipped (user-test-files-only).
• flake8 . --select=E9,F63,F7,F82: 0 issues.