bug: parallel find_all on empty result fails

[CoderJoshDK]

When you run a parallel version of the find_all, the pa.concat function fails: ArrowInvalid: Must pass at least one table

Instead, results needs to be validated that it has any data and if not, skip the concat.

[CoderJoshDK]

Another inconsistent thing of note; if the schema is different (similar to #371 ) but fully incompatible, it will fail but single threaded wont. For example if one value is a boolean and another is a string. It is late so I don't have an example for you right now. I am writing this for now as a note on the off chance I forget to come back to this.

For now, I have turned off parallel on my system. When a few (life) things slow down on my end, I might revisit here to get parallelism working for me again. (If you all don't get around to it)