A/B Test allocation logic may produce biased results

The way users are currently allocated to test cohorts using their MVT ID may lead to biased test results if there is more than one A/B test running at a time. Please see [this doc](https://docs.google.com/document/d/16CKxPaPV3bxCGygCH9pxXK6OS8ResfCavC-lpo5uAHc/edit?usp=sharing) for further information.

One potential solution to this problem would be to allocate users to test cohorts based on a seeded value, similar to the way we do in [support-frontend](https://github.com/guardian/support-frontend/blob/e4258f9e00d46797edba98f1d7b3347ae5196031/support-frontend/assets/helpers/abTests/abtest.ts#L356-L359).

Start with the linked A/B testing document, then locate the current MVT ID cohort-allocation logic in csnx. Compare it with the seeded-value approach shown in support-frontend's abtest.ts example. Done means concurrent A/B tests no longer produce biased cohort results, with the behavior verified by relevant tests.

Written by the indexing model from the issue text.