Mixed LF / CRLF causes confusion

I've now encountered a couple of GED files with mixed EOL. A good example is PDR.GED - the header has LF but the first record has CRLF.

The CRLF causes the parser to get confused and fails to recognise the record types. The output from the writer appends a NUL thanks to the CR.

see 01\PDR.ged

Start by reproducing the problem with 01\PDR.ged, comparing the LF header and CRLF first record described in the issue. Trace the parser and writer entry points to see how mixed EOL is handled; done means record types are recognized and the writer no longer appends a NUL for this input.

Written by the indexing model from the issue text.