bug: generate a valid class name when the project name starts with a digit

[Amadeus-22]

Closes #673.

• lean project-create --language python "2024 Momentum" wrote class 2024Momentum(QCAlgorithm):, which is a syntax error; the same happens for C#. The command still exited with code 0.
• convert_to_class_name now prefixes the name with an underscore when it is empty or starts with a digit (_2024Momentum). Other names are unchanged. The same helper is used by lean research.
• Added a parametrized test that creates the project and compiles the generated main.py. It fails on master and passes with the fix.
• pytest tests/commands/test_create_project.py tests/commands/test_research.py passes (30 tests, Python 3.12) and python static_analysis.py is clean.

🤖 Generated with Claude Code

[jhonabreul]

Thanks for the fix. I pushed a follow-up commit (e57b0ef) on top of it.

Names that still failed with the previous commit

The leading-digit case was fixed, but convert_to_class_name could still return a hard keyword. The command exited with code 0 and the generated main.py was a syntax error:

Project name	Class name	Why
none, true, false	None, True, False	capitalizing the word produces the keyword
_class, __import__, _lambda	class, import, lambda	the leading _ is capitalized as a no-op and then stripped, leaving the bare keyword

The second shape also produces C# keywords for C# projects (public class class : QCAlgorithm, which fails with CS1001). A fuzz over the characters the project-name validator allows, plus every entry of keyword.kwlist with each stripped prefix, found 187 breaking names.

What changed

• convert_to_class_name also prefixes _ when the stripped name starts with a lowercase letter or is a Python keyword. Keywords are lowercase in both languages except for None, True and False, so the lowercase check covers C# keywords without a separate list. _myproject now yields _myproject instead of Myproject; other names are unchanged.
• The parametrized test is now test_create_project_creates_valid_class_name_when_name_is_not_an_identifier and adds none, true, _class and __import__ to the existing digit cases; each one creates the project and compiles the generated main.py. The new cases fail on the previous commit. tests/commands/test_create_project.py and tests/commands/test_research.py pass (34 tests) and static_analysis.py is clean. The fuzz now reports zero breaking names.

A hyphen-prefixed name such as -def takes the same path as _def, but click rejects it as an unknown option before the command runs, so it is not part of the test.

[Amadeus-22]

Thanks, @jhonabreul. I read e57b0ef. I had only covered the leading digit and missed that capitalizing or stripping the prefix could still land on a keyword (none, _class). Using the lowercase check to cover the C# keywords without keeping a separate list is a neat way to handle both languages, and compiling the generated main.py in the test is a stronger check than the one I wrote. Nothing to add from my side.

[jhonabreul]

Thank you for your contribution!