Merging in changes to read and process sample3 (cs510) data

[lmilechin]

This PR makes a few small changes to ensure output data formatted like Sample3 can be read in and processed. This includes edits that allows for data with only one time step. This includes:

process2interp.m

• Passing the name of the subdirectory where the data exists rather than programmatically searching for it (this portion was unfriendly to the Sample3 format which included multiple subdirectories)
• Changing save location for precomputed interpolation coefficients- this way it can be saved without being overwritten
• Allowing data that has been pre-loaded to be passed in and then interpolated (passed in one time step at a time for now)

process2nctiles.m

• Change doClim to 0 to not do the monthly climatology (for now)

write2nctiles.m

• Handle single-timestep data. In this case, the dimlist includes time, dimsize does not. This produces an "Index exceeds matrix dimensions" later. Cleanest solution is to leave the time dimension off dimlist and dimname.

[gaelforget]

@lmilechin: As discussed, let’s avoid adding function arguments unless we want users to worry about them. For example:

[x] a hard coded iterateOverFiles=true in process2nctiles.m seems sufficient; not sure if there are cases where false is really preferable but my guess is that such instances are probably rare. [x] for latlon1D an automated switch would make the most sense I think; latlon1D=true would used whenever possible (i.e., whenever the lon and lat arrays can be simplified to 1D vectors) and otherwise set latlon1D=false so that 2D coordinates will be used (this approach seems to be less commonly supported by external software). Seems that this automation could take place within process2nctiles.m or maybe even within write2nctiles.m to minimize the use of function arguments. [x] for subDir, dirOut, filAvailDiag in process2interp please uncomment the previous approach which should remain as the default; then add code to deal with situations that require additional attention internally (i.e., not via function arguments) and interactively (if necessary when e.g. there is doubt). For example, if availablediagnostics.log is not present where expected by default then issue warning and prompt user to specify the directory where it can be found. Also please add documentation in the help section wrt expected file locations. [-] process2nctiles.m : adding doClim as an optional argument would be a clear improvement but let’s deal with clmbnds internally (i.e., not via function arguments) using the earlier code for setting clmbnds automatically by default unless a csv file that provides clmbnds is found in dirDiags. I feel that a good name for this file could be [‘clmbnds’ fileDiags ‘.csv’] or something like that.

Am happy to provide further clarification if needed.

[gaelforget]

@lmilechin The most recent commits addressed issues listed above (except for adding a doClim argument + file but this can wait). I updated the above comment accordingly.

A couple details:

• comment lines regarding iterateOverFiles that do not seem to apply anymore should be removed
• remaining / orphan references to latlon1D should be removed

A few more issues to sort out before merging:

• please undo changes made to read_nctiles.m (user can always read climatology_bounds separately)
• there seems to be an issue with the handling of tim
• dimension names need to be more specific