GC4 issues to fix

[ddeclerck]

This lists issues that came up when merging GC3 patches to GC4.

• [ ] 1. Decide what to do about gcdiff, espacially under MSVC (see here for details)
• [ ] 2. Correctly implement delay-loading under MSVC
• [ ] 3. Possibly drop usage of external cobxref (see here for details)
• [ ] 4. Fix any remaining failed test case (especially MSVC ones)
• [ ] 5. Rework the context-sensitive reserved words handling (or use a bigger type)
• [ ] 6. - Add --error-list -e to cobc
• [ ] 7. Investigate the two CHECKMEs about bdb_close_cursor in fbdb.c:ix_bdb_write_internal
• [ ] 8. Investigate the CHECKMEs in typeck.c:validate_move and in particular why it was necessary to add CB_LITERAL_P (checks to add in parser)
• [ ] 9. Investigate CHECKME about the need for comma in move.c:cob_move_display_to_edited (first in GC 3.x)
• [ ] 10. Investigate the CHECKME about moving cob_set_exception call in common.c:cob_module_global_enter
• [ ] 11. Investigate the TODO about cb_default_byte in codegen.c:output_initialize
• [ ] 12. Make status an enum instead of an int in fileio.c:cob_file_save_status
• [ ] 13. Move the IO status codes from common.h to fileio.h as per TODO
• [ ] 14. Make field type an enum instead of a short in common.h:cob_field_attr as per TODO
• [ ] 15. Investigate INITIALIZE BY VALUE (currently under #if 0) in field.c:validate_field_value
• [ ] 16. Check the code for setting last_exception_source (currently under #if 0) in common.c:cob_set_exception
• [ ] 17. Deal with the TODO about EOP exception in fileio.c:cob_file_save_status
• [ ] 18. Investigate the need to define READ_WRITE_NEEDS_FLUSH under MacOS ("LINE SEQUENTIAL COMMIT / ROLLBACK" test fails if not defined)
• [ ] 19. Check if should use strcpy or memcpy with computed max-length for file_open_name in fileio.c:cob_open
• [ ] 20. Check what we should do about the casts used to remove const on open_mode in fileio.c and others (eg. in cob_open_file)
• [ ] 21. Investigate failed manual tests "CRT STATUS clause" and "X/Open CRT STATUS clause" (see r4180)