[WIP] Silence userwarnings

[tv3141]

Closes #3078

• create IrisUserWarning and uses it where a UserWarning was raised before. This allows the silencing of all IrisUserWarnings from Iris without silencing UserWarnings from other libraries.
• silence all IrisUserWarnings (this can be made more specific)

[DPeterK]

Hi @tv3141 - thanks for this work to fix an outstanding Iris issue! Looks like some changes to master mean that your PR now has some merge conflicts; would you mind rebasing to fix these conflicts? Thanks!

[corinnebosley]

@tv3141 This PR still looks good to me, but we need you to rebase please. This isn't as straightforward as the last two that I've rebased for you; I'll need you to make sure the conflicts are resolved correctly.

[tv3141]

@corinnebosley @dkillick I will rebase this as soon as I can.

[tv3141]

This still needs a bit of work.

Also some tests fail because of ncdump. I do not know how my changes cause that yet.

ERROR: test_conflicting_global_attributes (iris.tests.test_netcdf.TestNetCDFSave)
----------------------------------------------------------------------
Traceback (most recent call last):
  File "/home/travis/miniconda/envs/test-environment/lib/python3.6/site-packages/scitools_iris-2.3.0.dev0-py3.6.egg/iris/tests/test_netcdf.py", line 779, in test_conflicting_global_attributes
    ('netcdf', 'netcdf_save_confl_global_attr.cdl'))
  File "/home/travis/miniconda/envs/test-environment/lib/python3.6/site-packages/scitools_iris-2.3.0.dev0-py3.6.egg/iris/tests/__init__.py", line 355, in assertCDL
    stderr=cdl_file, stdout=cdl_file)
  File "/home/travis/miniconda/envs/test-environment/lib/python3.6/subprocess.py", line 291, in check_call
    raise CalledProcessError(retcode, cmd)
subprocess.CalledProcessError: Command '['ncdump', '-h', '/tmp/tmpaa88ge1z.nc']' returned non-zero exit status 1.

[corinnebosley]

Yeah that's a bit worrying isn't it? I'll look into the testing environment; could be that something's been updated which has broken subprocess.

[corinnebosley]

On second thoughts I'll leave it a bit and have another look if and when that's the only test which fails.

[tv3141]

Yeah that's a bit worrying isn't it? I'll look into the testing environment; could be that something's been updated which has broken subprocess.

The problem was that the file did not exists. The root cause was missing indentation. I added #3219 to make this more verbose.

[tv3141]

So far I followed the warning docs on testing. This results in brittle tests that fail when the executed code creates FutureWarnings like we see them currently.
I think it is better to look for the warnings I want to test and re-emit all other warnings so that they are shown in the test log: 9eb61dd

[bjlittle]

@tv3141 Firstly, thanks for this awesome PR 👍

Sadly it's went rather stale. Also, we've decided to adopt logging as an alternative strategy, and reserve warnings for limited use throughout the code base. As such I'm going to close this PR, but reference it's efforts in an associated issue.

Thanks again