jobset.run was changed on 16 October 2015 in e4c35610
to return the number of failures and the result set. On 20 October 2016
in 5c691c63 jobset.run was changed to
allow jobs to be skipped, but the introduced code path only returned
the result set, giving jobset.run inconsistent return value semantics.
On 24 October 2016 7e9bd6ca introduced
a use of jobset.run's "skip_jobs" return value semantics, and on 8
March 2017 d159af31 resolved
jobset.run's inconsistent return value semantics but failed to correct
run_tests_matrix's use of the inconsistent "skip_jobs" return value
semantics.

Document status ordering rules

Sanitize before bigquery upload

Enhance bm_diff

Bumped version to 1.3.0-dev

Handle floats in bm_json

Change execution order of microbenchmarks

Finish streams that error. Don't histogram failed ops.

Give execute permission for Kokoro shell scripts

Fix a comment re wakeup_fd

Add test behavior to yapf_code.sh.

Error arena optimization

Add More Core Banned Functions

Add Zalloc to bq conversion table

Ensure GRPC_TIMER_END gets called

Dont use memory counters unless counters mode is on

Not calling this breaks our latency tracing/analysis code

Track zalloc times in microbenchmark

Simplify manual running of interop tests

Add global.json file to pick dotnet core SDK version

Server channel args

Add a test of creating a call on a lame channel in C++

Move to a customized benchmark diff tool

Add more jobs to Kokoro

Add copyright and license to bm_diff.py

Optimizing grpc_error Part 1

With this change yapf_code no longer changes the code under test when
executed from within the sanity tests.

Verbosely log flow control failures due to buggy implementations