call for volunteers to improve pbulk reporting

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

Currently, pbulk has suboptimal behavior under some error conditions:

  - It stops early if there is a python package that depends on another
    package that supports fewer versions, e.g.:

      pbulk-resolve: No match found for dependency py39-scipy>=0.19 of package py39-arviz-0.11.4

  - It notes on stdout if there is a bad bl3 include, the overall run
    continues, but does not list the package as failing, e.g.:

      make: "/usr/pkgsrc/test/test-dependencies/Makefile" line 56: Could not find ../../databases/3qlite3/buildlink3.mk
      make: Fatal errors encountered -- cannot continue
      . 1/1
      pbulk-scan: Scan failed for test/test-dependencies

  - When a DEPEND is wrong, there's an error, and then the package fails
    to build and is reported, but it's a little messy, e.g.:

      .............................Cannot change directory to "/usr/pkgsrc/fake/autoconf".
      .................. 47/47
      pbulk-scan: Scan failed for fake/autoconf

PMC would like pbulk to be improved so that:

  - The pbulk run as a whole continues even if any particular package is
    troubled.

  - The above errors are clearly reported in the pbulk output.

  - Packages that fail for any the above reasons are reported as and
    counted as failing.

  - Overall, a report that includes the rest of the pbulk run and the
    above issues is generated and can take the normal path to
    pkgsrc-bulk@.  Ideally fully in the machine-readable output, but
    incremental progress to list the above in the text output while
    allowing the run to continue is welcome incremental progress.
    
This is a call for volunteer(s) to prepare changes for pbulk to meet the
above goals.  (Given the number of changes and the likelihood of
multiple people, it seems best to have this in wip for testing and
review.)

Greg (on behalf of pkgsrc PMC)