Fix profiling scripts with a UTF-8 BOM

[mameikagou]

Fixes #1091.

Scalene reads the entry script as text, so a leading UTF-8 BOM reaches compile() and raises a SyntaxError. Use utf-8-sig to consume the BOM while preserving ordinary UTF-8 decoding.

Add real CLI regressions for scripts with a BOM, plain UTF-8, and non-ASCII UTF-8 content.

Validation on Python 3.12:

• The BOM script fails before the fix and executes successfully afterward.
• Eight focused tests pass (BOM/non-ASCII CLI regressions and the existing main-execution test); mypy scalene and ruff check scalene pass.
• The full test suite was not completed locally; the bounded run hit its time limit.