Improve inference in derivative/integrate

[jishnub]

On master

julia> p = fromroots(Polynomial, Float64[1,1,2])
Polynomial(-2.0 + 5.0*x - 4.0*x^2 + 1.0*x^3)

julia> using Test

julia> @inferred derivative(p)
ERROR: return type Polynomial{Float64, :x} does not match inferred return type Polynomial{Float64}
Stacktrace:
 [1] error(s::String)
   @ Base ./error.jl:35
 [2] top-level scope
   @ REPL[4]:1

julia> @inferred integrate(p)
ERROR: return type Polynomial{Float64, :x} does not match inferred return type Polynomial{Float64}
Stacktrace:
 [1] error(s::String)
   @ Base ./error.jl:35
 [2] top-level scope
   @ REPL[5]:1

This PR fixes these. The inference test is skipped for ImmutablePolynomial, as the length doesn't appear to be inferred (and I see several open issues regarding this, so I haven't bothered looking into it).

[codecov]

Codecov Report

Patch coverage: 100.00% and no project coverage change.

Comparison is base (2de280b) 79.78% compared to head (da9fe22) 79.78%.

Additional details and impacted files

@@           Coverage Diff           @@
##           master     #514   +/-   ##
=======================================
  Coverage   79.78%   79.78%           
=======================================
  Files          26       26           
  Lines        3216     3216           
=======================================
  Hits         2566     2566           
  Misses        650      650           

Impacted Files	Coverage Δ
src/polynomials/standard-basis.jl	80.36% <100.00%> (ø)

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[jishnub]

The invalidation test failure is because of an upstream (Pkg) issue

The test failure on v1.6 for ControlSystems is perhaps unrelated, and the repo doesn't seem to run CI tests on v1.6 anyway

[jverzani]

Thanks!

In #513 there are some timings where you can see the impact of this not being inferred. That PR addresses the ImmutablePolynomial case.