Remove pausing of data updates when running/queueing experiments

[mattseddon]

Related to #3178

Back in #1123 I added code to pause data updates whilst we queue experiments. This was because the commands used to update the data would take the read/write lock on the repository and cause the queueing process to fail. Now these commands are lockless so we don't need to pause updates anymore. The same thing happened intermittently when we started running experiments.

We need to do this to avoid trying to provide progress updates for batch-queueing experiments - e.g a grid search (see #3814)

Demo

Screen.Recording.2023-05-03.at.1.52.24.pm.mov

[shcheklein]

so, queueing is immediate? we were slowing it down artificially by running some commands?

[mattseddon]

so, queueing is immediate? we were slowing it down artificially by running some commands?

When we queued a big batch of experiments the functions that update the data underlying the table (that call exp show) were paused. We waited for the entire command to complete before unpausing them.

The same thing didn't happen for running experiments as they get processed asynchronously. For those we would start the experiment and then unpause updates.

[julieg18]

Looks good!

[codeclimate]

Code Climate has analyzed commit 2e2d864 and detected 0 issues on this pull request.

The test coverage on the diff in this pull request is 90.9% (85% is the threshold).

This pull request will bring the total coverage in the repository to 94.7% (0.0% change).

View more on Code Climate.