Avoid reading first chunk of DataChunkIterator on init if maxshape and dtype are specified

[rly]

If a DataChunkIterator is constructed with an iterator for data, then the first chunk is read on init in order to help determine the maxshape and dtype. This is unnecessary if both maxshape and dtype are provided to init, so we should avoid reading the first chunk. This helps with #136.

It looked like someone had already written in this condition years ago but left it commented out. This PR uncomments that condition.

[codecov]

Codecov Report

Merging #189 into dev will not change coverage.
The diff coverage is 90%.

@@           Coverage Diff           @@
##              dev     #189   +/-   ##
=======================================
  Coverage   70.32%   70.32%           
=======================================
  Files          30       30           
  Lines        5911     5911           
  Branches     1380     1381    +1     
=======================================
  Hits         4157     4157           
  Misses       1323     1323           
  Partials      431      431

Impacted Files	Coverage Δ
src/hdmf/data_utils.py	90% <90%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8ddae22...07d3b0c. Read the comment docs.

[ajtritt]

Seems like a good idea to bring that check back. Is there a reason why we don't move setting of maxshape to the getter i.e. only attempt to read the next chunk for setting maxshape when it is read ?

[rly]

Is there a reason why we don't move setting of maxshape to the getter i.e. only attempt to read the next chunk for setting maxshape when it is read ?

That seems like a reasonable thing to do. I'll refactor it.