improve performance in freqresp for tf

[baggepinnen]

This adds a custom method for transfer functions that avoid the matrix allocations from evalfr.
10x improvement in time for representable benchmark and 2000x in memory.

G = tf(1, [1, 1])
w = exp10.(LinRange(-2, 2, 200))
R = freqresp(G, w).parent
@btime ControlSystems.freqresp!($R, $G, $w);

julia> @btime ControlSystems.freqresp!($R, $G, $w);
  51.948 μs (2201 allocations: 146.94 KiB) # Before

julia> @btime ControlSystems.freqresp!($R, $G, $w);
  4.048 μs (1 allocation: 64 bytes)        # After

This PR also removes an extra specialization that was not needed.

[JuliaControlBot]

This is an automated message.
Plots were compared to references. 9/11 images have changed, see differences below.
After pulling this PR, please update the reference images by creating a PR to ControlExamplePlots.jl here.

Difference	Reference Image	New Image
✔️ 0.003
✔️ 0.003
✔️ 0.001
✔️ 0.001
✔️ 0.0
✔️ 0.0
❌ 0.042
✔️ 0.006
✔️ 0.011

[codecov]

Codecov Report

Merging #650 (fcf6dd5) into master (2f57fc9) will decrease coverage by 0.20%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master     #650      +/-   ##
==========================================
- Coverage   86.82%   86.62%   -0.21%     
==========================================
  Files          31       31              
  Lines        3309     3311       +2     
==========================================
- Hits         2873     2868       -5     
- Misses        436      443       +7     

Impacted Files	Coverage Δ
src/freqresp.jl	90.20% <100.00%> (-3.55%)	⬇️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 2f57fc9...fcf6dd5. Read the comment docs.