Bugfix libxml2 library location macOS, logging message for unit conversions, bugfix for metadata2html.py

[climbfuji]

This PR is associated and tested with the PRs listed below. For regression testing, see NCAR/ufs-weather-model#3.

Changes for ccpp-framework:

• scripts/mkcap.py: add logging message when performing unit conversions (addresses Missing log entry for automatic unit conversions #232)
• src/CMakeLists.txt: bugfix for macOS when specifying libxml2 libraries instead of auto-detecting them (use .dylib suffix instead of .so)
• scripts/metadata2html.py: update for out-of-source build changes in ufs-weather-model and gmtb-scm; for the ufs-weather-model, the documentation can be generated as follows:

ufs-weather-model$ # batch process for all schemes files in CCPP prebuild config
ufs-weather-model$ ./FV3/ccpp/framework/scripts/metadata2html.py -c FV3/ccpp/config/ccpp_prebuild_config.py -b FV3

ufs-weather-model$ #  for a single file
ufs-weather-model$ ./FV3/ccpp/framework/scripts/metadata2html.py -m FV3/ccpp/physics/physics/sascnvn.meta -o FV3/ccpp/physics/physics/

[climbfuji]

The output written to the ccpp_prebuild.py logs look like this:

INFO: Automatic unit conversion from m to um for effective_radius_of_stratiform_cloud_ice_particle_in_um before entering MODULE_mp_thompson SCHEME_mp_thompson SUBROUTINE_mp_thompson_run
INFO: Automatic unit conversion from m to um for effective_radius_of_stratiform_cloud_liquid_water_particle_in_um before entering MODULE_mp_thompson SCHEME_mp_thompson SUBROUTINE_mp_thompson_run
INFO: Automatic unit conversion from m to um for effective_radius_of_stratiform_cloud_snow_particle_in_um before entering MODULE_mp_thompson SCHEME_mp_thompson SUBROUTINE_mp_thompson_run

[climbfuji]

Associated PRs:
#241
NCAR/ccpp-physics#355
NCAR/fv3atm#3
NCAR/ufs-weather-model#3

[grantfirl]

Approved.