Parse doxygen tag files

[mkitti]

I parsed the Doxygen tag file and generated a tab separated value file, hdf5_func_urls.tsv

The next step is to use these URLs when generating the documentation.

[simonbyrne]

Haha, I had been working on this as well. I'll open a draft PR with what I have, maybe we can combine.

[mkitti]

I'm still interested in taking some ideas from #1044 such as replacing the for loop with a xpath query, but I"m running short on time.

[mkitti]

We still need to modify h5doc: https://github.com/JuliaIO/HDF5.jl/blob/ffc0b66cdab5be3c8e29a637c4514c97d7e126d8/src/HDF5.jl#L65

[mkitti]

I changed this to use the HDF5 v1.14 documentation rather than develop to narrow the gap with #1044.

I'm going to merge this over #1044. My main reasons for preferring this over #1044 are as follows.

• Here in 1043, the documentation URLs are statically embedded statically at generate time within the documentation rather than at parse time via macro that does that a xpath lookup.
• The generation step here does a single pass over the XML tag file to generate a tsv file that is easily converted into a Julia Dict. The xpath lookup ends up multiple passes over the XML file.

I do want to revisit some of the ideas in #1044 later, especially with a pure Julia XML parser.

The current errors are due HDF5_jll binaries losing SZIP support. That should be fixed shortly with HDF5 1.14.2.