regression reporter: keep some objects in storage for the whole lifetime of the regression

**As a regression reporter, I want to be able to access all the information and build objects of a regression regardless of when it was first reported so I can test it and investigate it**

## Background

Currently KernelCI saves the build objects and logs of a regression only for a couple of weeks. If someone wants to study a regression that happened earlier than that there'll be a lot of information missing. The reporter will need to re-create the build objects but there'll be no way to tell if the test run will be exactly the same as the original one.

For each regression, KernelCI keeps the history of test runs related to it, since the last one that passed until the most recent one that failed. It'd be enough to keep the objects of the last passed run and the first failed run (and, optionally the latest one as well). Once the regression is resolved and fixed definitely, they can be safely removed.

Start by tracing the regression reporter's handling of regression history, test runs, build objects, and logs, then identify where the current couple-of-weeks retention is enforced. Done means retaining the last passed run and first failed run for each regression, with optional retention of the latest run, and removing them after the regression is definitively resolved.

Written by the indexing model from the issue text.