Add a benchmark for img_as

[hmaarrfk]

Description

I included naive implementations below. They probably aren't the best, but they would be what I would use if I were to roll my own.

xref: #4268

Checklist

• Docstrings for all functions
• Gallery example in ./doc/examples (new features only)
• Benchmark in ./benchmarks, if your changes aren't covered by an
  existing benchmark
• Unit tests
• Clean style in the spirit of PEP8

For reviewers

• Check that the PR title is short, concise, and will make sense 1 year
  later.
• Check that new functions are imported in corresponding __init__.py.
• Check that new features, API changes, and deprecations are mentioned in
  doc/release/release_dev.rst.
• Consider backporting the PR with @meeseeksdev backport to v0.14.x

[pep8speaks]

Hello @hmaarrfk! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

• In the file benchmarks/benchmark_img_as.py:

Line 65:27: E221 multiple spaces before operator

Comment last updated at 2019-11-04 13:18:59 UTC

[sciunto]

@hmaarrfk You included notebook, i guess it is unattended.

[sciunto]

I'm just worried that one function will fail (without testing). Besides that, I'm +1.

[sciunto]

I looked at this function np.iinfo and looks like it is working only with interger dtypes. In skimage.util.dtype, we also have the range specifications. Perhaps you can use that.

[hmaarrfk]

dtype_out is guaranteed to be an integer at this point.

I guess I'm trying to replicate the rint function. I've tested this internally, and it works quite well. I used linspace with about 1E6 points to see how the two compare for uint16 data types.

[hmaarrfk]

benchmark coverage was proposed in #3329

Probably needs a rebase and moving it to Azure?