add tensorflow as the KERAS_BACKEND for qkerasV3 tests

[jmitrevs]

Description

This should fix the qkeras3 pytests issues introduced with qkeras_v3==1.2

Type of change

QKeras v3 1.2 needs to have the KERAS_BACKEND environment variable set, so this sets it in the CI.

• Bug fix (non-breaking change that fixes an issue)
• Other (Specify) - CI

Tests

This is a CI fix so it will only be visible in the CI. Compare with the qkeras3 pytest in #1497.

Checklist

• I have read the guidelines for contributing.
• I have commented my code, particularly in hard-to-understand areas.
• I have made corresponding changes to the documentation.
• My changes generate no new warnings.
• I have installed and run pre-commit on the files I edited or added.
• I have added tests that prove my fix is effective or that my feature works.

[calad0i]

Sounds like a bug from upstream in qkeras v3 to me. In keras v3, if KERAS_BACKEND is not set, it shall be regarded as tensorflow.

[jmitrevs]

However, we need a workaround till it is fixed. It seems better to explicitly set it vs to lock the version to an older version of qkeras3.

[calad0i]

I would do version lock in this case

[jmitrevs]

I don't have a strong preference either way. Let's see what others say. I don't see a particular downside in setting the KERAS_BACKEND explicitly.

[makoeppel]

I can set the backend default to tensorflow in qkerasv3.

[makoeppel]

I fixed the default backend to tensorflow in qkerasv3.

[jmitrevs]

So I think this PR is no longer necessary. The latest version of QKeras3 no longer fails in this case. It is kind of odd, though, that currently pquant tests use the template_keras3_backend in the CI while keras3 tests do not. Feel free to close, though.