GitLab

In fact, it's not unitargets itself that is slow, it is
`Collection::find_all_by` (called via `get_my_targets`) as collections are
not indexed.

'source_dataflow_rule'==undef can only be matched by the targets created
right before since they will either be given a real source_dataflow_rule or
be forgotten below once we've identified the correct DataflowRule

This brings a minor performance improvement.

This saves us from parsing the method name again and again. This
command:
```
generate_graph.pl -pipeconfig Bio::EnsEMBL::Compara::PipeConfig::EBI::Ensembl::ProteinTrees_conf -output /tmp/dummy.png -mlss_id 34
```
now takes ~10 seconds instead of ~22 seconds by saving >3.8M calls to
`parse_undescored_id_name`. To be clear the 3.8M calls still happen but
they're directly routed to the new subs instead of being processed by
AUTOLOAD.

Since we can't easily match the host name, check the permissions on *all* the hosts and tell whether there is write access to _one_ of them

* version/2.4:
  Copyright update
  Copyright update
  Copyright update
  Copyright update
  Copyright update
  Copyright update
  bugfix: the / is a remnant of the previous regex
  Show the stack trace if $self->param() is called without arguments
  bugfix: there can't be any whitespace in the pipeline name

* version/2.3:
  Copyright update
  Copyright update
  Copyright update
  Copyright update
  Copyright update

documentation update: remove instructions to turn on respecialization when running beekeeper, as it is now on by default

* version/2.2:
  Copyright update
  Copyright update
  Copyright update
  Copyright update

* version/2.1:
  Copyright update
  Copyright update
  Copyright update

* version/2.0:
  Copyright update
  Copyright update

* version/1.9:
  Copyright update

This is especially the case of Compara, not much eHive.