Non-ASCII end of line characters in commit messages break git scanning (JVM 1.7 fixes)

If a Unicode end of line character appears in a commit message (i.e. 

…

…U+2028, U+2029 or U+0085) the state machine which parses the output of 'git whatchanged' gets confused.

A typical log message is:

2012-06-04 09:42:40,089 ERROR - Unexpected commit statement C:1cc26fc19e8efd2875b0337592c3cf8ce786384c while in state HEADER, existing commit 4f9e156f645ef9983c380299a8714bbbba8b76cb

The fix is to recognise Unicode end of line characters while scanning git output.

This might also affect hg repos, but this is unknown. The fix to this issue will ensure that hg is not affected.