Our generated IR is often not valid for llc/lli/clang

Describe the bug The generate IR will often fail if executed by llc, lli or clang, this doesn't always cause issues but sometimes the compiler just segfaults (on oscat for example). We can add a module.validate() right after generation to prevent that

The main issue with fixing this right now is that a lot of tests might already fail, but I think it's important to get a valid IR when generating.

Below is a list of issues detected by our tests when activating validate():

[x] #559 This was found in the case_with_enum_expressions_in_case_selectors test, we can solve the test, but we also need validation to not allow duplicate cases (in this case, DOWN is BASE * 2 same as FORWARD * 2
[x] #562 It happens often when VAR_OUTPUT are involved (optional_output_assignment)
[x] #569 Seen in typed_enums_are_generated, mainly an issue with enums
[x] #570 It seems we don't convert both types to DINT in comparisons : aliased_ranged_number_type_comparing_test
[x] #571
[x] #572 Needs Investigation related to sized variadics. `test_max_int, also occurs in by ref functions
[x] #573
[x] #584

The following tests in correctness still need investigation, they are probably similar to the cases above:

[x] correctness::custom_datatypes::using_duplicate_enums_with_casts
[x] correctness::functions::by_ref_and_by_val_mixed_function_call
[x] correctness::functions::optional_output_assignment
[x] correctness::functions::optional_output_assignment_in_functions
[x] correctness::pointers::binary_expressions_for_pointers
[x] correctness::pointers::binary_expressions_for_pointers_with_function_return
[ ] correctness::strings::string_as_function_parameters
[ ] correctness::strings::string_as_function_parameters_cast
[ ] correctness::strings::wstring_as_function_parameters
[ ] correctness::strings::wstring_as_function_parameters_cast
[x] #625
[ ] correctness::functions::by_ref_and_by_val_mixed_function_call

PLC-lang / rusty

Our generated IR is often not valid for llc/lli/clang #529