DistributedClosestPoint should gracefully deal with variations allowed by mesh blueprint

[gunney1]

DistributedClosestPoint is limited in what the exact form of incoming data. Ideally, it should be able to handle any valid blueprint mesh. At a minimum, it should detect valid data that it can't handle and gracefully exit with a useful diagnostic message. Currently, it is limited to double reals and integer integral data and may not exit gracefully if that is not the case.

This was brought up in a project meeting. It is related to issue #835.

[rhornung67]

More detailed group discussion about what we want/need is needed for this.

[rhornung67]

@BradWhitlock mentioned Conduit topology iteration as a possible model to consider as a starting point.