`TfrType` assumptions and conversions

[aphalo]

Was: T2A() and normalise() errors and missing Afr2A() method.

T2A() seems to not work correctly when Tfr == 0 and Rfr != 0. The problem could be caused by a wrongly set TfrType attribute in the data I was using, or could be an error in the computations. In the first case a validity test would be missing and in the second case, a plain bug.

The missing Afr2A() would have been useful for testing and diagnosing the problem, and adding it should be the first step.

[aphalo]

I found one cause of spurious normalization-warning messages due to an optimization used mainly in 'ggspectra'. It is now fixed. Also added support for collections of spectra to normalization-related functions. Afr2A() remains to be implemented. Testing is not yet done. I need to check that the bug raised in this issue is actually fixed and add a unit test.

• [x] Check that unit tests cover all conversions
• [ ] Ensure that unit tests actually test the copying and "edition" of metadata correctly and fully during conversions.
• [ ] Add Afr2A() methods.

[aphalo]

Main problem seems to be the poorly documented and possibly inconsistently implemented assumption that A and Afr are always defined as "internal" irrespective of the TfrType setting of the filter_spct object.

• [x] document what the assumptions are
• [x] function convertTfrType() needs to be fixed to make use of the Rfr.constant stored in the filter.properties attribute.
• [x] methods A2T() and T2A() need to be tested with and without Rfr.constant available in the filter.properties attribute.

[aphalo]

This is for the most part done. As there are some pending taks, I push the Milestone to the next one 0.11.2.