Follow-up from "MFDv5 / XLALCWMakeFakeData: fix SFT band selection when loading frames"

The following discussion from !1582 (merged) should be addressed:

@david-keitel: I've done this now, using lalapps_splitSFTs on the original SFT to get the comparison basis. This fails now with a 1-bin offset at the end: fmin=299 and Band=10 => MFDv5 includes the 309 bin, splitSFTs does not! I could work around this here by extending the splitSFTs band, but I wonder if that's a rounding convention difference we need to track down...