Remove LZ4HDF5 from blosc subcompressor list

[jwlodek]

I missed this on review of the LZ4HDF5 PR. The compressBlosc function, uses the enum value for NDCodecBloscComp_t to get the compressor name from the bloscCompName array. Following this addition, the codecs after lz4 no longer matched up. Specifying a zstd codec would actually cause a segfault since it was last. - I noticed that when I added the NDPluginCodec unit tests, but I thought my version of blosc just didn't support zstd.

My understanding is that the lz4hdf5 codec is standalone and not meant as a blosc subcompressor, is that correct? If it should exist as a subcompressor, then it'll need to be added to the string array and the record as well.

[jwlodek]

Also, I wonder if this enumeration (and the array of matching string names) should move to Codec.h alongside the main compressor enumeration - I have a driver that is producing blosc compressed arrays, but to include this enumeration I need to include the NDPluginCodec header, which doesn't seem like it should be the case.

[MarkRivers]

Also, I wonder if this enumeration (and the array of matching string names) should move to Codec.h alongside the main compressor enumeration - I have a driver that is producing blosc compressed arrays, but to include this enumeration I need to include the NDPluginCodec header, which doesn't seem like it should be the case.

Sorry about that mistake in added lz4hdf5 to blosc enum.

I agree that moving the blosc enum to Codec.h is a good idea.

Should we also rename Codec.h to NDCodec.h, so it is less likely to conflict with other packages?

[jwlodek]

Should we also rename Codec.h to NDCodec.h, so it is less likely to conflict with other packages?

I think that is also a good idea. I'll make both changes and add them to this PR. I'll also check which drivers import Codec.h and update those as well.

[jwlodek]

I've renamed the file as well as the name of the Codec_t struct and the codecName array to be prefixed with ND. I also moved the static name arrays into a C++ file w/ extern, since using static would lead to compiler warnings about the array of blosc compressor names being unused in most of the files that imported this header. I also changed that to an array of C++ std::string to follow the pattern set by the base NDCodecName array.

[jwlodek]

It doesn't seem like any existing drivers actually import this header - in ADEiger the codec name is set directly as a string without the array lookup. So I think this change would actually be non-breaking for existing drivers.

[MarkRivers]

It would probably be good to change ADEiger to use the array lookup to future proof it.