Improve idmaker name selection in case of duplicate ids in parametrize

[palaviv]

As @RonnyPfannschmidt mentioned in pull request #1468 the name selection in case of duplicate ids had some problems.
Let's take for example:

import pytest
@pytest.mark.parametrize("a", [1,2,3,4,5,6], ids=["a", "a", "b", "c", "b", "l"])
def test_a(a,b):
    pass

Before this change the tests names would be:

• test_a[0a]
• test_a[1a]
• test_a[2b]
• test_a[3c]
• test_a[4b]
• test_a[5l]

After this change:

• test_a[a0]
• test_a[a1]
• test_a[b0]
• test_a[c]
• test_a[b1]
• test_a[l]

[nicoddemus]

Please include a "thanks palaviv" and mention #1474. You might want to do the same for the item just above this one. 😉

[palaviv]

It seems that the travis-ci failed because of some error unrelated to the tests:

The command "curl -s -o python-3.5.tar.bz2 https://s3.amazonaws.com/travis-python archives/binaries/$(lsb_release -is | tr "A-Z" "a-z")/$(lsb_release -rs)/$(uname -m)/python-3.5.tar.bz2" failed and exited with 18 during .

Is there a way to rerun the build?

[The-Compiler]

I just restarted the failing build.

[nicoddemus]

Thanks again @palaviv! 😄