Use fastcov to speed up coverage for CI

[ruffsl]

Make use of fastcov to speed up CI:
https://github.com/RPGillespie6/fastcov

[ruffsl]

A rather nice improvement... Not too shabby :P

Before: 31m 25s

https://app.circleci.com/pipelines/github/ros-planning/navigation2/5455/workflows/625353dc-ec89-41b3-a554-4cb2f6dcecf2/jobs/19971/parallel-runs/0/steps/0-116

After: 34s

https://app.circleci.com/pipelines/github/ros-planning/navigation2/5458/workflows/f65c0be8-e5c3-46ec-8a5b-d20e726f6f48/jobs/19985/parallel-runs/0/steps/0-117

[SteveMacenski]

When did that start taking 30 minutes? Looking back only a month it was taking about 2 minutes https://app.circleci.com/pipelines/github/ros-planning/navigation2/5301/workflows/1131a535-ce5b-4bc2-bd26-0b3af0a0836f/jobs/19546/steps

[ruffsl]

With the gap in CI runs from the galactic breakages, I could only narrow down the regression between theses workflow runs:

• Before regression:
  • https://app.circleci.com/pipelines/github/ros-planning/navigation2/5364/workflows/37e3f336-bec9-4250-916c-ae6a9178c1c3/jobs/19764
• After regression
  • https://app.circleci.com/pipelines/github/ros-planning/navigation2/5448/workflows/289167a7-9002-451d-a8e5-86696b4b4b08/jobs/19943/steps

I'm not sure what changed here though. Any guesses? Some of my guesses:

• Performance regression upstream in lcov package
• Increase or change in how we generate coverage with gcc

The coverage mixin hasn't changed for more than a year, so the cmake and python args wouldn't have changed.

https://github.com/colcon/colcon-mixin-repository/blob/master/coverage.mixin

[SteveMacenski]

ros2/rclcpp@0251f70 its also the date rclcpp released 11.0.0, so maybe something in the ROS2 stack was updated and got back up to us in compile flags?

Only other thing I could think of it this commit #2330 did something odd, but I don't see how that's possible

I'm good with this change either way once you are

[ruffsl]

I'm good with this change either way once you are

Codcov doesn't show any change in coverage for this PR, so I'll go ahead and merge.

https://codecov.io/github/ros-planning/navigation2/commit/dbc8153498f562961106f8876f5e88756b1e4ba7

I'm also going to open a ticket with colcon-lcov-result so ask if they've noticed the same performance regression, but at least this should resolve our issue in the meantime.