bug in upsample_and_fit_scatter_estimate for cylindrical

[KrisThielemans]

Current code reads

STIR/src/scatter_buildblock/upsample_and_fit_scatter_estimate.cxx

Lines 62 to 73 in 11b7c9b

	ProjDataInMemory interpolated_direct_scatter(emission_proj_data.get_exam_info_sptr(),
	interpolated_direct_scatter_proj_data_info_sptr);
	bool actual_remove_interleaving = remove_interleaving;
	if (remove_interleaving && emission_proj_data.get_proj_data_info_sptr()->get_scanner_sptr()->get_scanner_geometry()!="Cylindrical")
	{
	warning("upsample_and_fit_scatter_estimate: forcing remove_interleaving to false as non-cylindrical projdata");
	actual_remove_interleaving = false;
	}
	if (emission_proj_data.get_proj_data_info_sptr()->get_scanner_sptr()->get_scanner_geometry()=="Cylindrical")
	interpolated_direct_scatter_proj_data_info_sptr->reduce_segment_range(0,0);

that seems incorrect, as the interpolated_direct_scatter_proj_data_info_sptr is reduced in size after creating interpolated_direct_scatter. This happened at
1b128f2 part of #1006. @danieldeidda we had a conversation in that PR about it, but in the end left it as-is.

Original code was

STIR/src/scatter_buildblock/upsample_and_fit_scatter_estimate.cxx

Lines 55 to 61 in aaa0f8e

	shared_ptr<ProjDataInfo>
	interpolated_direct_scatter_proj_data_info_sptr(emission_proj_data.get_proj_data_info_sptr()->clone());
	interpolated_direct_scatter_proj_data_info_sptr->reduce_segment_range(0,0);
	info("upsample_and_fit_scatter_estimate: Interpolating scatter estimate to size of emission data");
	ProjDataInMemory interpolated_direct_scatter(emission_proj_data.get_exam_info_sptr(),
	interpolated_direct_scatter_proj_data_info_sptr);

Found while debugging the TOF branch with @NicoleJurjew

[KrisThielemans]

In fact, I think we should reduce the segment range to 0,0 also for blocks. Currently, we are calling interpolate_projdata (whcih handles segment 0 only), and inverse_SSRB in all cases. Therefore, allocating interpolated_direct_scatter with only the 0 segment is clearer and will save memory.

(TBH, it is not clear to me why @NicoleJurjew 's modification of putting a TOF-loop in interpolate_projdata caused a crash if we don't fix this issue, as it first sight it should just create a "too large" interpolated_direct_scatter, but still work I guess. In any case, the current code is inefficient).