Test-centile_bands

[eatyourpeas]

Overview

This pulls in changes in the python package that corrects the centile_bands in the Measurement class on adding CDC reference.
In the process it was necessary to allow centile values more than a decimal place, since CDC BMI has complicated maths for BMI values > 99, where they are on the 99.9th or 99.99th centiles. Rounding here would have returned 100. This broke the tests so these have been fixed.

While doing this I noticed that one of the request params for fictional-child-data is reference. This is unnecessary and could lead to errors since fictional-child-data is already prefixed by the reference. If a user had requested UK-WHO children against the CDC endpoint they might have received the wrong children back.

This is now fixed but by removing reference from the request schema this will make a breaking change. I don't think anyone uses this endpoint apart from us but we have a postman suite which would break, the openapi spec would need refreshing too.
Otherwise all the tests are updated and passing.

@mbarton - how should we handle this merge?

Code changes

Please describe the changes you made here.

Documentation changes (done or required as a result of this PR)

Please describe any changes to documentation here.

Related Issues

List any issues related to this PR here.

Mentions

@mentions of the person or team responsible for reviewing proposed changes.

[eatyourpeas]

Just to add for comment really
The tests fail because the final digit (16th) for me locally does not match what is calculated on running the same tests on github. I am not going to get too lost in this level of precision, so inspite of what i said above, I have pruned the return values in the CDC and UK-WHO responses for centile values to 4 dp.