[PATCH Rebreather] Notification: Changes to sample structure

[Willem Ferguson]

Below is a patch that has been applied to the rebreather branch. This is 
a first step
in preparing for code changes to accommodate rebreathers.
It is not submitted for pushing to master but the patch message is 
circulated here for comment.

     Change sample structure for typed variables and RAM space reduction

     In line with previous discussions, the structure of sample in dive.h
     is changed as follows: some new typedefs:
     1) A type of o2pressure_t is created with units .mbar.
         This affects the variable po2.
     2) A type of shortduration_t is created with units .seconds.
         This affects variables ndl and stoptime.
     3) A type of bearing_t is created with units .degrees.
         This affects variable bearing.
     4) The other integer variables are more strongly defined
         as uint8_t (variables sensor, cns and heartbeat)

     The way that these variables now need to addressed in the rest
     of the source code, e.g:
     sample.po2  now becomes  sample.po2.mbar
     sample.ndl becomes sample.ndl.seconds

     Because of these needs, many places in the existing code base
     need changing. In this patch the following files are affected:

     dive h: functions: sampe, typedefs
     dive.c: functions: merge_samples, fixup_pressure
     divelist.c: functions: calculate_otu, calculate_cns, add_dive_to_deco
     libdivecomputer.c: function: sample_cb
     load-git.c: functions: parse_sample_keyvalue, get_shortduration, 
get_o2pressure
     parse-xml.c: functions: sample_start, sample_end, 
shearwater_profile_sample
     planner.c: functions: tissue_at_end, create_dive_from_plan, plan
     profile.c: functions: populate_plot_entries
     save-git.c: functions: save_git, sample
     save-xml.c: save_sample

The methodology was to address compilation errors that arose after changing
the sample structure in dive.h. In most cases it was a mechanical 
replacement
e.g.  sample-po2  changed to sample.po2.mbar; but this was not always
the case.

I checked easily checkable things like the xml export, but this cannot 
claim to
be a conclusive test and that other bugs have not been introduced.

Kind regards,
willemf