Fixes to EventIDRange
This MR is a proposed solution to @ssnyder's comments on MR !831 (merged).
- Fix buggy comparison
- Constructor of EventIDRange sets the starting time-stamp to 0 if a run/lumi-indexed EventIDBase is provided or the Run/Lumi to 0 if a timestamp-based EventIDBase is proveded. This way, ranges that are soly limited by time-stamps have infinte range in terms of run/lumi (analogous for soly run/lumi-limited ranges) and can be narrowed down by subsequent intersections with other ranges.