If ``getTimestamps`` was used with a frame range not starting from 1, it
would return a vector with 0's. For example:
```
getTimestamps(1,1, 3:5) = [0 0 3 4 5]
```

This commit fixes the indexing issue that causes this.

The last commit addressed an issue when a file was missing timestamp
information. The code will now output a warning + a timestamp vector of
zeros. However, the timestamp units were not being assigned in this
case.

In this commit, tsunits are now assigned to a blank string ''. This
should avoid issues with code compatibility when the timestamp
information is missing.

When timestamps were not saved in the file metadata, getTimestamps
would return a "reference to non-struct array" error. I added some
statements which would check that the call to getPlaneDeltaT is not
empty. If it is, the code will return a warning along with a timestamp
vector which is all zeros.

Issue: bfReader.getTimestamps(1, 'channel') was giving a java out of index error.

Cause: The metadata.getPlaneDeltaT function requires both series index and image index to be 0 based.

Solution: Subtracted the image plane index (``bfIndex``) by 1 in getTimestamps().

The TimeRange input validation was initially isinf(x) || isnumeric(x). However, entering a range (e.g. 1:10), the validation function returns an error because isinf(x) is a vector but || is a scalar operation.

Corrected this by changing to all(isinf(x)) ...

Merge branch 'dev'

* Added cleanup function to call close() on bfObj

* Removed stacking functions (MAP, zStack etc.)
* Unified the interface i.e. getPlane adn getTimestamp now uses (Z,C,T) coordinates. getXYPlane uses (C, T, SiteIndex) [Z coordinate is unsupported for this function because I do not have an example to test on].
* Added an example ND2 so that users can verify their installation. The example.nd2 is from Joe 4x4 binned image IQBio 2016

Updated getXYloc to fix iSeries (was zero based, but now it should be 1-based. Also updated the getPlane function call to match the signature of the planeSpec.

* Made a change to the way the plane specification was working in getPlane. Previously, when using getTile, the planeSpec would be binned into a single vector - this then causes the inputParser to fail because the value is no longer scalar. Now the planeSpec is coverted first into a cell, then put into inputParser which handles it correctly.

* Also updated the testBioformatsImage to check that getTile is correct.

Fixed bug with getTile. The tile height was using image width and therefore was not being calculated correctly.

Moved unit string parser to static methods. Updated PxUnitStr and timestamps (timestamps now has a second output with the unit str)