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

changed selenocystein implementation to use SeqEdits and be more consistant with Transcript implementatin

for now edits are disabled by default, b/c will slow down dumping, also flush TranscriptMapper when enabling/disabling edits

changed transcript edit implementation to use SeqEdits and made it possible to enable/disable editing

supercontig length should be max(superctg_end) not length because occasionally supercontigs do not start at 1

moved snp typing methods into utility class in an attempt to make very large transcript module more manageable

added a caching algorithm to somewhat improve efficiency of certain usage patterns (such as transcript/prediction transcript dumps

made fetch_all_by_PredictionTranscript method more efficient (if exons are not in high-level coord system)

So when there is a problem of the auto-disconnect it prints out the
offending query to facilitate finding the missing $sth->finish.

'invalidates 1 active statement handle' warning

'invalidates 1 active statement handle' warning

Rather than calling base class DBI::st DESTROY explicitly, rebless statement handle and let perl call it automatically. - this is better b/c DBI::st does not always define a DESTROY method (suggested by jgrg)

Added automatic copy of schema_description.html to private-web/htdocs/Docs to save doing it manually

Changed output filenames to match what is used in the static web module - removes need to rename when updating