Remove global Flint* objects

[lgoettgens]

As suggested by @fieker in oscar-system/Oscar.jl#1379 (comment).

I would consider this a breaking change for Nemo, and this possibly needs some changes downstream in Hecke and Oscar for uses of FlintZZ and FlintQQ.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 87.49%. Comparing base (0271fc0) to head (312639d).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1935      +/-   ##
==========================================
- Coverage   87.51%   87.49%   -0.02%     
==========================================
  Files          97       97              
  Lines       35532    35538       +6     
==========================================
- Hits        31096    31095       -1     
- Misses       4436     4443       +7     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[lgoettgens]

Triage decided (ref oscar-system/Oscar.jl#1379 (comment)) to first replace all uses of these downstream, then wait for some time, and only then merge this. If we need deprecations or not may be discussed then.

[thofma]

can we keep the definitions but not export them? they are kind of handy

[lgoettgens]

can we keep the definitions but not export them? they are kind of handy

the gaussian ones or the QQ/ZZ? And what is the benefit of e.g. FlintQQ over QQ?
(just trying to understand your use-case)

[thofma]

The Gaussian ones

[lgoettgens]

We could just add QQi (instead of FlintQQi) as an unexported global for this (and similarly for ZZi)

[thofma]

also works for me

[fingolfin]

As a milder step I would suggest that as a milder change we instead (or at first) modify Oscar and/or Hecke to not import resp. not re-export these. That would then perhaps be "breaking" for Oscar and/or Hecke but not Nemo, which arguably affects fewer people... But I am open to this PR here as well shrug