[STABrAPP] - Bug when fitting model on a variable with CV=0 or all NA

[jframi]

In case a variable has a CV equal to 0 or only missing values in a given Environment (BrAPI study), it is included in the model, resulting either in:

• Error message "could not fit model on [Environment name]"
• rows with no data in heritabilities column and CV=0 in the Model statistics table
  This is also causing the "Download report" button to fail

[jframi]

In addition to filtering data for Variables with CV=0 or missing data only before running model, it is necessary to check that the STA object return by the call to fitTD contains valid models. In some case model cannot be fitted because of too many NAs or any other reason and the $mFix and $mRand of the returned STA objects are NULL.

[jframi]

Would be useful to warn the user that some combinations of variable/environment were discarded from the analysis