Prevent name conflicts with internally used identifiers

[treiher]

During code generation but also in other circumstances there is a risk of name conflicts between internally used identifiers and user-defined identifiers. A solution would be to prevent a user from defining identifiers with the prefix RFLX_, as it is already done for package identifiers.

Known problematic cases:

• Unknown enum literals in PyRFLX https://github.com/Componolit/RecordFlux/blob/dc4e51978521b9fb9f1f5bbe10eebfe02447cf82/rflx/pyrflx/typevalue.py#L255
• Variable declarations in sessions
• Name of session allocator package can conflict with package for session, message or sequence

[treiher]

The issue gets even more difficult when warnings on hiding are activated (-gnatwh is currently not used due to Componolit/Workarounds#27). Scalar types defined in a RecordFlux specification will conflict with parameter names of generated SPARK subprograms (e.g., Ctx or Buffer). IMHO prefixing all parameters with RFLX_ will impair the readability of the generated code too much, maybe a shorter prefix would be acceptable, e.g., P_ for all parameters. We are already using F_ for the literals representing message field names. So one solution could be to generally prohibit identifiers beginning with [A-Z]_ in specifications.