Case found where Compass Component migration tool is NOT idepotent, will fail to update most components

[jtcsProjHub]

Although the README file indicates that multiple executions of the migration script will not create multiple Compass Components of the same name, it does in fact do just that.

Additionally, of the 95 components it created in Compass, the first pass of the script was only able to successfully replace 6 of those components within the Jira Work Items for the target project. I reran the script an hour later just in case there was some backend Atlassian database caching issue or something (like you see with JSM Asset updates), and the script:

1. Created a duplicate Compass Component for most of the items. I honestly don't know what the logic was behind what was duplicated and what wasn't. All of the Compass Components existed following the first run.
2. Associated 14 components to the project instead of the 6 it did the first time
3. Those 14 components didn't completely include the ones from the first run of the script

I'll look through the script in more detail to see if I can figure anything out. Changes that I had made before running it:

1. Changed the component type to OTHER
2. Added a team ID for ownership to the GraphQL

[jtcsProjHub]

Found the issue. The function find_compass_component_in_jira seems to make the assumption that:

1. You don't have a lot of identically named legacy components across projects
2. If one of the components in your system with the name being queried is a Compass component, the it will be the first component returned by the API

Since many of our components are cross-project in some way, we had many instances of legacy components with the same name in Jira. There also didn't seem to be any rhyme or reason to what order the API call returned its results. I fixed the issue by changing line 131 to the following:
filtered_response = [component for component in response['values'] if (component['name'] == component_name and 'ari' in component)]

I also removed the now redundant 'ari' check from line 132, but did keep return statement on line 135 as-is in case there happen to be multiple existing Compass components with the same name. For that situation, it seems reasonable to just pick the first one.

[jtcsProjHub]

Reopening the issue so it shows up on the list of open issues for better visibility to others. Will see if I'm allowed to submit a PR for it as well in a bit.