fix: add another illegal id

[pauldambra]

If we query

select team_id, any(distinct_id), any(JSONExtractRaw(properties, 'distinct_id'))
from events 
where timestamp >= now() - interval 4 day
and JSONExtractRaw(properties, 'distinct_id') = '"null"'
group by team_id

(while investigating something else

We note that some teams send "null" as distinct id (i.e. double quote null double quote as a string)

We shouldn't allow this

[tiina303]

don't we want to do all of these single or double quoted?

[pauldambra]

I wondered about that... looking in the DB we only see them double quoted, but I guess that doesn't mean we'd never see them single quoted 🤔

[pauldambra]

testing for every illegal id showed that even though we check for the empty string you couldn't get into this branch of the code if the anon distinct id was the empty string

[tiina303]

we would exit https://github.com/PostHog/posthog/pull/17325/files#diff-c51e19d65898f9cd1b4aaad32e5163e7b44507026ce49301507a2cbd6fe5ccdbR306

[tiina303]

I'd update the tests & there's a failure to look into

[pauldambra]

@tiina303 I updated the tests... and it turns out we make assumptions about casing that clashed with my change. E.g. we do allow Nan as an illegal id.

Decided to unwind my casing change rather than go deeper into the rabbit hole. If you're happy with that I'll merge this later on today 👍

[tiina303]

E.g. we do allow Nan as an illegal id.

why? Should we?