Detection of duplicate aspects

[treiher]

The specification of duplicate aspects shouldn't be accepted (e.g., for First and Size aspects for message fields and Readable and Writable aspects for channels):

package Test is

   type T is mod 2**8;

   type M is
      message
         A : T;
         B : Opaque
            with First => 123, Size => 123, First => A'First, Size => A'Size;
      end message;

   generic
      C : Channel with Readable, Writable, Readable;
   session S with
      Initial => I,
      Final => F
   is
   begin
      state I
      is
         M : Test::M;
      begin
         C'Read (M);
      transition
         then F
      end I;

      state F is null state;
   end S;

end Test;

[treiher]

The Parser in rflx/specification/parser.py is responsible for parsing a specification and creating a model out of it. For the actual parsing the langkit-based parser is used. The returned AST is evaluated to create a model for all specified types and sessions.

The processing of field aspects (like First and Size) is done here:

https://github.com/Componolit/RecordFlux/blob/227113adac75c458074d76bc4d55f28f508b4d50/rflx/specification/parser.py#L983-L1002

The processing of channel aspects can be found here:

https://github.com/Componolit/RecordFlux/blob/227113adac75c458074d76bc4d55f28f508b4d50/rflx/specification/parser.py#L538-L544

@mhadhbir These are probably the two places where modifications are needed to fix the two problems mentioned in the issue description. The error reporting should be done using an error parameter of type RecordFluxError, as it is done in similar functions. To make the task a bit more challenging, please also have a look at other uses of aspects. For example, messages can have Checksum aspects and transitions in sessions can have Desc aspects. Maybe we have the same issue there. The right place for adding tests is tests/unit/specification/parser_test.py. You can use the existing test_parse_error_* test functions as a guideline.