bumps tensorflow to >=1.15.1,<2.0.0

[lsh-0]

regarding the security issue GHSA-977j-xj7q-2jr9

[lsh-0]

We are additionally releasing TensorFlow 1.15.1 and 2.0.1 with this vulnerability patched.

I guess they haven't yet:

[2020-01-30T06:39:34.696Z] Collecting tensorflow<2.0.0,>=1.15.1 (from -r /opt/sciencebeam-trainer-delft/requirements.cpu.txt (line 1))
[2020-01-30T06:39:34.696Z]   Could not find a version that satisfies the requirement tensorflow<2.0.0,>=1.15.1 (from -r /opt/sciencebeam-trainer-delft/requirements.cpu.txt (line 1)) (from versions: 1.13.0rc1, 1.13.0rc2, 1.13.1, 1.13.2, 1.14.0rc0, 1.14.0rc1, 1.14.0, 2.0.0a0, 2.0.0b0, 2.0.0b1)
[2020-01-30T06:39:34.696Z] No matching distribution found for tensorflow<2.0.0,>=1.15.1 (from -r /opt/sciencebeam-trainer-delft/requirements.cpu.txt (line 1))

[lsh-0]

hrm, no, they're definitely there.

[lsh-0]

@de-code could you take a look at this please? it's not the version of pip being used. It might be related to this issue.

[de-code]

I added pip to the requirements.build.txt which I already had in the project but wasn't used by the Dockerfile. I guess it was either setuptools or wheel that fixed it.
(Adding it to requirements.build.txt also has the added advantage that it works with a virtual environment without using docker)

[lsh-0]

thanks @de-code , it's appreciated