Optimize map_coefficients, change_base_ring for some inputs

[fingolfin]

Namely if the input polynomial is a QQMPolyRingElem and the transformation "function" is ZZ resp. a fpField.

Before this patch:

julia> Qxy,(x,y) = QQ[:x,:y]; Zxy,_ = ZZ[:u,:v];

julia> f = 11*(x^2/2 + ZZ(3)^50*x*y^5/7 + y^6/12);

julia> @b $f*84
160.000 ns (4 allocs: 200 bytes)

julia> @b change_base_ring($ZZ, $f*84)
1.559 μs (45 allocs: 1.633 KiB)

julia> @b change_base_ring($ZZ, $f*84; parent = $Zxy)
1.002 μs (24 allocs: 808 bytes)

julia> @b map_coefficients($ZZ, $f*84; parent = $Zxy)
996.680 ns (24 allocs: 808 bytes)

After this patch:

julia> Qxy,(x,y) = QQ[:x,:y]; Zxy,_ = ZZ[:u,:v];

julia> f = 11*(x^2/2 + ZZ(3)^50*x*y^5/7 + y^6/12);

julia> @b $f*84
111.693 ns (4 allocs: 200 bytes)

julia> @b change_base_ring($ZZ, $f*84)
733.806 ns (30 allocs: 1.227 KiB)

julia> @b change_base_ring($ZZ, $f*84; parent = $Zxy)
242.117 ns (9 allocs: 392 bytes)

julia> @b map_coefficients($ZZ, $f*84; parent = $Zxy)
249.607 ns (9 allocs: 392 bytes)

As a bonus we now also have a special mul! method which is even faster but is not suitable for the faint of heart:

julia> @b mul!($Zxy(), $f, 84)
129.041 ns (5 allocs: 192 bytes)

---

This needs work, at least tests. But I didn't want to forget this, plus this way I can get feedback.

[codecov]

Codecov Report

Attention: Patch coverage is 94.59459% with 2 lines in your changes missing coverage. Please review.

Project coverage is 88.28%. Comparing base (3b6fb3f) to head (deff7a7).
Report is 3 commits behind head on master.

Files with missing lines	Patch %	Lines
src/flint/fmpq_mpoly.jl	94.59%	2 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1994      +/-   ##
==========================================
+ Coverage   88.24%   88.28%   +0.03%     
==========================================
  Files         100      100              
  Lines       36740    36767      +27     
==========================================
+ Hits        32422    32459      +37     
+ Misses       4318     4308      -10     

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

[fingolfin]

This should be ready for review now

[fingolfin]

These field sizes were wrong; it didn't matter for us in practice but it showed up when I used dump to understand why sort_terms! didn't do what I thought it would...

julia> dump(polynomial_ring(QQ, [:u,:v]; internal_ordering = :degrevlex)[1]; maxdepth=1)
QQMPolyRing
  nvars: Int64 2
  nfields: Int32 3
  ord: Int64 4294967298
  deg: Int32 1
  rev: Int32 0
  lut: NTuple{64, Int64}
  lut1: NTuple{64, UInt8}
  S: Array{Symbol}((2,))
  __attrs: #undef

[fingolfin]

I addressed all issues with this PR I was/am aware of