Enable more Windows performance optimizations as in Linux

[adrpo]

Currently we have less optimizations in Windows:

• GC parallel mark is disabled ​https://github.com/ivmai/bdwgc/issues/119
• parallel parsing of library files is disabled (see #3805)
• parallel code generation is disabled for Windows
• we handle reading of text files in ANTLR3 differently ​https://github.com/OpenModelica/OMCompiler-3rdParty/blob/master/antlr/3.2/libantlr3c-3.2/src/antlr3filestream.c#L107

These should all be fixed if possible.

[adrpo]

I'm about to fix all of these issues except "parallel code generation" with:
PR957: ​https://github.com/OpenModelica/OMCompiler/pull/957
PR9: ​https://github.com/OpenModelica/OMCompiler-3rdParty/pull/9
I'm not sure it will work so we'll see after the Windows job finishes.

[adrpo]

Fixed with 250e6b/OMCompiler-3rdParty and 7e1286/OMCompiler.
Parallel code generation seems to be already available, so no need to do anything there.

We need a bit more testing with the Windows job and with new OMEdit from a nightly-build to make sure everything works fine.

[adrpo]

Tested OMEdit myself and it works rather well.
Windows build has no different failures than before the changes.
I will close this ticket.

[sjoelund.se]

Milestone deleted