Add workaround for JISC newspapers

[thobson88]

The JISC newspapers dataset, after pre-processing with the jisc-wrangler tool, exhibit a slightly different directory structure to that assumed by alto2txt.

Instead of a single newspaper issue subdirectory of the form mmdd/ the JISC structure separates the month & day into separate subdirectories: mm/dd/.

This results in warnings like:

extract_text.xml_to_text:19683:WARNING:Unexpected directory: 02

and the XML files in the subdirectory are not processed.

The task is to implement a simple workaround to accommodate the non-standard JISC directory structure.

[thobson88]

Fix added on branch 24-jisc-fix-catch.

I've tested this successfully on a single JISC title but have not yet tested on non-JISC data (to make sure the fix hasn't introduced any bugs).

[kallewesterling]

This looks related to #30?