GitLab

* server: fix unneeded model reloads when setting `OLLAMA_NUM_PARALLEL`

* remove whitespace change

* undo some changes