Renumseq

This is going to be a VERY handy new option, one that I am planning on using a lot. Instead of making a whole new command, say `renameseq` it seemed better to just add an option to `renumseq` as it's possible that these two steps (of renaming and renumbering) might want to be happening at the same time anyway.

From the `--help` output:


--rename NEW_SEQNAME Rename the DESCRIPTIVE_NAME part of SEQ from its existing name to
NEW_SEQNAME. When using this option then the command will exit with an
error unless exactly one SEQ is being renamed and/or renumbered.

This new option is non-trivial to implement as there are lots of ways for it to be misused, probably by accident by the user. I expect over the next little while that a few bug-fix-point-releases may be coming to catch the odd weird case not caught in this new-feature-release.

However, this release is extensively tested, and lots of possible sources of user-errors are attempted to be caught before execution. As such, there are plenty of good error messages printed before exiting the program when errors are found. Also, the EXIT-status is set so it can be checked in the shell.


EXIT_NO_ERROR = 0 Clean exit.
EXIT_ERROR = 1 Internal error other than argparse - currently not used.
EXIT_ARGPARSE_ERROR = 2 The default code that argparse exits with if bad option.

--touch` option now working correctly in all cases.

In the prior release (v1.2.2) `--touch <DATE>` wasn't working properly as the date wasn't being parsed or used, and in this case the mod/access times were being set to to June 1, 2001 no matter what date was passed as an argument (due to a line of test code that never got fleshed out). I expect no one was using this feature so wasn't reported as a bug.

`--touch` with no arguments was working properly in setting the dates to the current time.

Omitting `--touch` entirely was also working properly and the mod/access times of the files being renumbered was NOT changed, like the `mv` command which does not affect those times.

So now `--touch [CC]YYMMDD[-hh[mm[ss]]]` correctly sets the access/mod times of the files being renumbered to the date specified. Note that zero-padding on MM DD hh mm ss is enforced to avoid ambiguity in parsing the date.

see `--help` for details

When using --start N, such that N would generate a negative offset for renumbering the frames,
the code generated a type error, since a -1.0 was being multiplied against a frame offset,
which made it a float, but later in the code, a reformat string (which only got used when using --start)
was expecting an int. Simply fixed the multiplication to use -1 instead of -1.0.

The regular expression that was used to match lsseq-native-format SEQuences didn't catch the case when there could be a dot in the file-extension. Eg. cache sequences could have this, like 'bgeo.gz'.

Also cleaner interpretation of --silent flag to only suppress errors and warnings (not --verbose output).

Decided to kill the --fixBadPadding option, it was a poorly conceived, overcomplicated in how it was going to integrate and furthermore, wasn't yet implemented. It's gone now.

It was a solution to a rare problem. Instead I decided to make a new util called fixseqpadding to handle the rare times that this fix needs to happen in keeping with unix philosopy, make tools do simple things well and chain them together. Seems a far better approach.

Also, much better README.md now and better --help output. Some other logic cleanup in the util as well. Passes all regression tests.

'renum' was far too generic, plus my other commands all have 'seq' in the names, so it's more consistent with them. Better for pypi.org too. renum was taken, and my renum was too specific to take the general sounding name 'renum' out of the running.

Also, this is a full rebuild of the package, much easier to install now with pip.