This came out of BF-39015 since more tests have been added to the search e2e testing suites. Because tests run concurrently in evergreen, multiple search indexes can be built on the same mongot at the same time, which leads to OOMs showing up quite a bit on memory-intensive variants (e.g. AUBSAN). 

SERVER-109341 will try to fix this by reducing the number of concurrent jobs and increasing the size of the machine, but we may need a more permanent solution:  

1) It would be ideal if we could do something similar to SERVER-104913 for the hybrid search tests – we create one, smaller, generic search index for all tests that require it to avoid OOMing issues in the e2e suites. 

2) If it seems too difficult to refactor the hybrid search tests to use shared indexes, we could start by using less dimensions for the indexes in the tests.