Querystring data is often corrupt

[colmsnowplow]

While testing the http sink, I found that many GET requests fail validation in micro due to the cx or ue_px fields in the querystring. For some examples, I found that when I decoded the b64, it was a self-describing json which ended with a trailing character, like this: ...}7.

36 of 50 get requests fail in this way.

Suspect that something is going wrong in the generation of the data itself, or in the b64 encoding of the data. It's possible also that it is to do with the encoding of the querystring itself, but this has occurred for two different ways of implementing the querystring, and manually curling the exact string produced by querystring.toString() results in the same thing.

Since our priority is collector testing, and this issue only becomes problematic downstream of that - at enrich - and since it may be time consuming to figure it out, for now I am deprioritising this behind collector-related P1 objectives