highlighting bulk-test in pbulk reports, and #failures?

[Greg Troxel <gdt%lexort.com@localhost>]

In scanning bulk build reports sent to pkgsrc-bulk, I find that it is
very good at showing the package breaking the most other packages, but
not so good at other aspects of situational awareness.

I'd like to see a few changes in the standard reports and wonder if
anybody else agrees and especially if someone wants to implement them.

  For packages that failed because of a dependency, show a count of how
  many dependencies failed, vs just that it failed.  Now, it's hard to
  separate bulk-test-large with 1 failure from 50.

  Add a section which lists all packages matching meta-packages/bulk-*
  alphabetically, showing the # of dependency failures or SUCCESS.  It's
  ok that bulk-large depends on medium and that if medium fails those
  counts aren't attributed to large.


This could look like (this is actual data from
  https://bulktracker.appspot.com/build/ag1zfmJ1bGt0cmFja2VychILEgVidWlsZBiAgICzkv6bCww/meta-pkgs/
).

===============================================================================

Bulk test results

Package                          Failed Deps
-------------------------------------------------------------------------
bulk-large			           3
bulk-medium                          [indirect-prefailed per bulktracker*]
bulk-small                           SUCCESS
bulk-test-boost                            1
bulk-test-essential                  SUCCESS
bulk-test-go                               1
bulk-test-llvm                             7

===============================================================================

and be sorted just before or after "packages breaking the most" and
definitely before "build failures".

Right now it is more effort to tell if a bulk-foo has or has not
succeeded.

*From bulk-tracker, I could not figure out why bulk-medium failed.