Naming scheme for channel ranges

[tjgalvin]

At the moment I am using the following naming scheme to indicate channel ranges when forming images with only a subset of channels:

SB59058.RACS_1626-84.beam18.round4.i.ch96-97

This would be supplied to the imager (wsclean -name) and output products would be formed around this prefix (e.g. the ImageSet will have its prefix set to this).

Two problems:

• the scheme itself may not be the best. These should at the very least be zero-padded. There is a question around whether we want a range, or simply have the starting index. My feeling is the channel range may be useful in the future should we want to image multiple channels at a time, instead of a single. Same thought around doing the time-range imaging
• The regex's that are used to separate the wsclean added naming components, and the general flint name parser both need to be expanded to recognise this chNNNN-NNNN field, should it be present. I believe it is this point that is creating files with names like:

SB59058.RACS_1626-84.beam18.round4.i.ch96-97-image.optimal.image.fits

in the spectral line workflow

[tjgalvin]

@AlecThomson may have strong feelings on point 1.

[tjgalvin]

Addressed as part of #196