gcpy/plot/core.py
- Moved function "gprofng_text_to_data_units" (from module
  "gprofng_functions") to this module as "text_to_data_units".
  This will make it available as a general plotting function.

gcpy/profile/gprofng_functions.py
- Import "text_to_data_units" from "gcpy.plot.core"
- Fixed incorrect error message string

gcpy/profile/vtune_plot_hotspots
- Added this function to create a bargraph from Intel VTune
  hotspot reports.

CHANGELOG.md
- Updated accordingly

Signed-off-by: Bob Yantosca <yantosca@seas.harvard.edu>

docs/source/Code-Profiling.rst
- Added a section describing how to plot hotspots with
  gcpy.profile.vtune_plot_hotspots.

CHANGELOG.md
- Updated accordingly

Signed-off-by: Bob Yantosca <yantosca@seas.harvard.edu>

gcpy/profile/vtune_plot_hotspots.py
- Skip plotting entries such as "wait.h", "mutex.c", etc, as
  these can have multiple entries in an Intel VTune report, which
  will confuse the bargraph plotting.  We mostly care about user
  routines that can be modified to save time, rather than system
  calls.

Signed-off-by: Bob Yantosca <yantosca@seas.harvard.edu>

gcpy/benchmark/modules/benchmark_scrape_gcclassic_timers.py
- Updated the marker that denotes when to stop reading from a
  GEOS-Chem Classic log file from "----------" to "-"*78 (that is,
  78 dashes in a row).  This is necessary in order to prevent text
  appended by Intel VTune to the end of a GEOS-Chem log file from
  interfering with the scraping of the timers.

CHANGELOG.md
- Updated accordingly

Signed-off-by: Bob Yantosca <yantosca@seas.harvard.edu>

This merge brings PR # (Update the "end of search" criteria in
benchmark_scrape_classic_timers.py to avoid being spoofed by
output from Intel VTune, by @yantosca) into the GCPy development
stream.

In gcpy/benchmark/modules/benchmark_scrape_gcclassic.timers.py, the
code starts reading the timers information backwards from the end of the
file. The signal to stop reading was if a line contained "----------".
However, when output from the Intel VTune profiler is appended to the
end of a GEOS-Chem Classic log file, there is a line with at least
as many "-" characters, which causes the parsing of the file to end before
the timing information is reached.

The solution is to change the criteria to look for a line with 78 "-"
characters in a row (which occurs just before GEOS-Chem timer row. This
will allow the parsing of the log file to proceed normally.

Signed-off-by: Bob Yantosca <yantosca@seas.harvard.edu>

This merge brings PR #364 (Add script to plot hotspots from Intel VTune
reports, by @yantosca) into the GCPy development stream.

PR #364 does the following:

1. Adds gcpy/profile/vtune_plot_hotspots.py, which can create
   bargraph plots (similar to gprofng_functions.py) from hotspot reports
   from Intel VTune. This tool accepts hotspot reports listed by function
   and by line.

2. Moves the gprofng_text_to_data_units functions out of
   gcpy/profile/gprofng_functions.py to gcpy/plot/core.py, so that it can
   be used as a generic plotting function.

3. Updates ReadTheDocs documentation in docs/source/Code-Profiling.rst
   accordingly.

Signed-off-by: Bob Yantosca <yantosca@seas.harvard.edu>

This merge brings PR #367 (Fix grid area calculation #367,
by @1Dandan) into the GCPy 1.6.3 development stream.

PR #367 fixes an issue with the computation of grid areas for
GCHP cubed-sphere and stretched grids.

Signed-off-by: Bob Yantosca <yantosca@seas.harvard.edu>

CHANGELOG.md
- Removed duplicated entries for version 1.6.2.  These were
  probably a result of Git automatic merges.

Signed-off-by: Bob Yantosca <yantosca@seas.harvard.edu>