Use a single process for mining commits in the HTTP service

Since we need to mine only a few commits per call, it doesn't make sense
to pay the overhead of starting many threads and processes.

This should help with #1374

Author: marco-c

http_service/bugbug_http/boot.py

@@ -45,7 +45,9 @@ def boot_worker():
 
     # Download commits DB
     logger.info("Downloading commits DB...")
-    commits_db_downloaded = db.download(repository.COMMITS_DB, support_files_too=True)
+    commits_db_downloaded = db.download(
+        repository.COMMITS_DB, support_files_too=True, use_single_process=True
+    )
     if not ALLOW_MISSING_MODELS:
         assert commits_db_downloaded
 

http_service/bugbug_http/models.py

@@ -167,7 +167,9 @@ def schedule_tests(branch, rev):
     )
 
     # Analyze patches.
-    commits = repository.download_commits(REPO_DIR, rev_start=first_rev, save=False)
+    commits = repository.download_commits(
+        REPO_DIR, rev_start=first_rev, save=False, use_single_process=True
+    )
 
     commit_data = commit_features.merge_commits(commits)