Speed up PixelConditionsSummaryTool/SCT_ByteStreamErrorsTool at high thread count

The PixelConditionsSummaryTool & SCT_ByteStreamErrorsTool currently use tool-wide mutexes to protect SG::SlotSpecificObjs. While this can be important (if slots < threads, possibly multiple threads would try to enter the critical section at once), the current setup forces serialization of (expensive) cache-filling for every thread, which for large #slots results in an enormous delay in the first event. This MR makes slot-specific mutexes instead, which lock at the necessary granularity for each slot to execute independently.