Rolling back the code to 1.20 because of issues with re-binning the human Y chromsome density values resulting in a -ve value.

production database.

Many improvements to reduce memory usage. Now uses registry, so don't need to specify databases explicitly, just use e.g. -species human

Fixed bug that showed up on Y chromosome and problem with not deleting entries when no analysis description was present, see release 58 notes for more details.

  my ( $current_start, $current_end );
as these variables were being declared twice

process the first 100 of them.

in human is number 118 in length, no gene, %gc, or SNP density features
were being calculated for it.

reinstated.  The analysis_id in the analysis table and in the
analysis_description table seems to become unsynced otherwise.

store each density feature as it is created rather than building a list per seq_region - smaller memory footprint (but repeats still leaking mem)