investigate coalescing first and second phase of indexing

[vladak]

As identified when indexing AOPS, the first phase of indexing (when the history cache is generated) can take a long time due to one of the repositories having much bigger revision/commit count than the rest of the repositories which leads to under-utilization of the CPU. It basically leads to a long tail of the first phase. The second phase of the indexing can currently start only after the first phase is completed. To speed up the overall indexing time, some thought should be given to whether it is possible to coalesce the first and second phase (on per repository level).