*Really* fix --long file decode, plus other bits

[christopherkobayashi]

It turns out there was an off-by-one in the record seek logic that fouled up --long decodes.

I've also turned on stack backtrace for the parser, so that we can debug this:

Traceback (most recent call last):
  File "/home/wileyc/bin/xbas99.py", line 574, in <module>
    status = main()
  File "/home/wileyc/bin/xbas99.py", line 531, in main
    data = program.getSource()
  File "/home/wileyc/bin/xbas99.py", line 266, in getSource
    istext = (lit[0] in "ABCDEFGHIJKLMNOPQRSTUVWXYZ" +
TypeError: 'NoneType' object has no attribute '__getitem__'

... there's an invalid opcode (0xff) creeping in somewhere, and I think it's an end-of-record marker that isn't getting purged by the long-to-short routine.

Finally, I turned on autodetect for long and TIFILES format files based on first few characters of the file.

[endlos99]

Just wanted to let you know that I'm really busy with the FRC right now. Once I have time for xdt99 your PR will be first on my list.

[christopherkobayashi]

No worries, I understand. Thanks for taking the time to update me.

-- Chris
GPG key fingerprint 603A 9892 C54E 0D77 0F0F D88A B061 38E3 7BF0 7BE8

[christopherkobayashi]

DSDD80T appears to work now, without regression for other formats.