Fix FFM backend on Windows

[Glavo]

The FFM backend appears to be untested and will crash when using it, I'm trying to solve these problems.

• Avoid lookup ioctl and isatty on Windows;
• Make SYMBOL_LOOKUP initialized earlier to avoid NPE
• Fix wrong alignment;
• Load Kernel32;
• Fix incorrect type conversion;
• Impl CLibrary for Windows;
• Fix string encoding;

There are more issues waiting to be fixed, WIP.

[Glavo]

It no longer crashes, but some features still don't work.

[Glavo]

It mostly works now, but I suspect there are hidden bugs that will need to be fixed.

[Glavo]

This PR is ready for review.

@gnodet I think it would be better to move AnsiConsoleSupport and org.fusesource.jansi.ffm to org.fusesource.jansi.internal?

They are the same low-level API as the existing CLibrary/Kernel32 and do not seem to be suitable as part of the public API.

I'd like to do some refactoring before they are officially released, what do you think?

[gnodet]

This PR is ready for review.

@gnodet I think it would be better to move AnsiConsoleSupport and org.fusesource.jansi.ffm to org.fusesource.jansi.internal?

They are the same low-level API as the existing CLibrary/Kernel32 and do not seem to be suitable as part of the public API.

I'd like to do some refactoring before they are officially released, what do you think?

Yes, that makes sense.

[Glavo]

This PR is ready for review.
@gnodet I think it would be better to move AnsiConsoleSupport and org.fusesource.jansi.ffm to org.fusesource.jansi.internal?
They are the same low-level API as the existing CLibrary/Kernel32 and do not seem to be suitable as part of the public API.
I'd like to do some refactoring before they are officially released, what do you think?

Yes, that makes sense.

Okay, when this PR is merged, I will do this work in a new PR.