AMIGAOS: Mutliple fixes

[raziel-]

• Add FS directory creation
• Remove hardcoded newlib toolchain (there may be follow-up FR's)

[bluegr]

I don't understand this comment. Is there a possibility that a node with _bIsDirectory = true is not a directory?

[sev-]

Could you please fix the commit log messages to have the proper "AMIGAOS: " prefixes?

[raziel-]

@sev-

I have no idea how to do that from github web page.
I can close and redo the PR?

[digitall]

@raziel- , @sev- : Should be fixed now. I pushed to fix the commit prefix and squash the trivial typo commit into the previous commit.

[raziel-]

@bluegr

Regarding the comment...this will be revised.

@capehill will (hopefully) join and help out, please stand by

[capehill]

@raziel- and other devs

Is it so that before calling backend's createDirectory, FSNode actually checks the existence and type: https://github.com/scummvm/scummvm/blob/master/common/fs.cpp#L156

In that case, backend could be simplified to something like:

bool AmigaOSFilesystemNode::createDirectory() { 
    BPTR lock = IDOS->CreateDir(_sPath.c_str()); 
    if (lock) { 
        IDOS->UnLock(lock); 
        debug("AmigaOSFilesystemNode::createDirectory() -> Directory '%s' created", _sPath.c_str()); 
        _bIsValid = true; 
        _bIsDirectory = true; 
    } else { 
        warning("AmigaOSFilesystemNode::createDirectory() -> Failed to create '%s'", _sPath.c_str()); 
    }     
     
    return _bIsValid && _bIsDirectory; 
}

This is still untested.

[raziel-]

Would that be applicable?

Tested locally and working

[bluegr]

This looks good now, and is aligned with what other platforms do in the same function. Since the changes are working for you @raziel, they should be fine to merge.

Thanks for your work!