more updates

[jbathegit]

Part of #254
Part of #328
Part of #445

[jbathegit]

OK all good now. There was a bug in rewnbf that was always restoring the subset pointer to point to the first subset (i.e. the beginning) of a restored BUFR message whenever the subroutine was called with isr=1, even if the subset pointer had been pointing to a different subset in the message when the routine was called previously with isr=0. I had fixed this bug in an earlier commit, but there was an existing Python test which had been overcoming this bug by re-reading up to the expected subset in the message. So once I fixed the bug in the library, that Python test began failing until I now just removed the extra script lines which had previously been re-positioning the subset pointer to overcome the bug.