Misc fixes pointed out by Eclipse PyDev static code analysis

[cdeil]

Here's fixes for a few scipy issues pointed out by my editor (Eclipse PyDev).
I'm not familiar with the code and didn't test, so please check the changes carefully.

There's two more that I think might be real issues where I don't know what the correct fix is:

Description Resource    Path    Location    Type
Undefined variable: get_psolve  iterative.py    /scipy/scipy/sparse/linalg/isolve   line 198    PyDev Problem
Undefined variable: ct  hb.py   /scipy/scipy/io/harwell_boeing  line 257    PyDev Problem

https://coveralls.io/files/59355318#L199
https://coveralls.io/files/59355112#L258

All of these issues are not covered by unit tests.

[coveralls]

Coverage remained the same when pulling c68403e on cdeil:cleanup into ce9d9d1 on scipy:master.

[larsmans]

I guess a unit test for this change would be in order.

[rgommers]

Thanks @cdeil, those fixes all look good. Unit tests for the kdtree and fitpack changes would indeed be very helpful as @larsmans points out.

[rgommers]

From a quick check it also looks like the node comparisons are missing from ckdtree.

[cdeil]

@larsmans @rgommers These issues were just pointed out by a static code analysis ... I've never used these parts of scipy and it would take me quite some time to figure out how to add the unit tests you suggest, which I don't have at the moment.
Maybe you can quickly add the extra unit tests or ping the maintainers of the relevant modules to do it?

[larsmans]

I've sent to a PR to @cdeil's repo with a test for the KDTree stuff.

[cdeil]

Thanks @larsmans ... merged your PR into this one.

[coveralls]

Coverage remained the same when pulling c4e0be7 on cdeil:cleanup into ce9d9d1 on scipy:master.

[rgommers]

Added dblint test in 860bf8e and merged this PR in e572003. Thanks @cdeil and @larsmans.