Fail asv workflows when performance deteriorates

Nobody has claimed this yet.

Fail the ASV workflows if at least one of benchmarks deteriorates above a certain threshold (e.g. 150% / 200%).

**asv-main.yml**:
- Execute `asv run` without `|| true`, allowing the command to exit with a non-zero return code
- Submit benchmarks and deploy dashboard on gh-pages, if the previous command succeeded or not

**asv-pr.yml**:
- Execute `asv continuous` without `|| true`, allowing the command to exit with a non-zero return code
- Output the performance comment on the PR if the comparison was successful or not

**asv-nightly.yml**
- Execute `asv continuous` without `|| true`, allowing the command to exit with a non-zero return code
- Write the commit hash to cache, if the previous if the comparison was successful or not

More information about "on failure" executions [here](https://docs.github.com/en/actions/learn-github-actions/expressions#status-check-functions).

Start with .github/workflows/asv-main.yml, asv-pr.yml, and asv-nightly.yml, and inspect how each currently handles asv run or asv continuous failures. Update the workflow failure conditions while preserving the requested submit, dashboard, PR comment, and cache steps, then verify the workflows still perform those steps when comparisons fail.

Written by the indexing model from the issue text.