Changes

clean up some debugging output in the smokedb api emulation
add /api/nntp_from_id/someid and /api/nntp_data/someid for internal transfers to the dev db without loading the real nntp server, though it's available for others to use too

NNTP Reports received by email and posted to the site are again parsed immediately on receipt, as with the old site.

only colour the status with the status rather than the whole row, since it was a bit overpowering
remove the sha column, it was only useful for the commit rows and was otherwise a waste of space. The commit subject is now a link to the commit.
sort smokes by os, host, id, previously it was only os. I wanted to group the duplicate smoke reports (between DB and NNTP reports), this isn't perfect, but it's better than just os
clean up the compiler name specifically to work around NNTP report parsing. This will be removed once NNTP report parsing is re-worked, which is waiting on the dev database being separate from the live database which depends on this iteration.
ported several of the back end tools to the new config system
add filtering to the recent commits page

fix presentation of multiple commit sequences, this was presenting all of the commits in each group rather than just the commits for that group
when stripping headers the second raw line of a header would always be stripped (typical for Subject:)

branches that have been rebased or squashed would often mix commits from before and after the rebase/squash, these should now be distinguished

rewrite using Mojo/DBIx::Class (rather than the horrible CGI/pure Mason)
do my own SmokeDB report presentation for speed
start saving log contents and do my own SmokeDB log presentation for speed
make the filters more obvious (after feedback from khw)
hovering over report lines now highlights the line to make it easier to distinguish that line from others
filter out most headers when presenting nntp reports