rm matvec & furthur optimize performance

[Roger-luo]

this further optimizes the performance of rotation gates.

@benchmark apply!(r, $(put(5, 2=>Rx(0.5)))) setup=(r=rand_state(5))

before:

BenchmarkTools.Trial: 
  memory estimate:  688 bytes
  allocs estimate:  17
  --------------
  minimum time:     424.719 ns (0.00% GC)
  median time:      441.246 ns (0.00% GC)
  mean time:        473.802 ns (5.99% GC)
  maximum time:     10.010 μs (95.16% GC)
  --------------
  samples:          10000
  evals/sample:     199

after:

BenchmarkTools.Trial: 
  memory estimate:  32 bytes
  allocs estimate:  2
  --------------
  minimum time:     72.515 ns (0.00% GC)
  median time:      92.314 ns (0.00% GC)
  mean time:        94.292 ns (0.72% GC)
  maximum time:     1.508 μs (93.35% GC)
  --------------
  samples:          10000
  evals/sample:     954

[codecov]

Codecov Report

Merging #71 into master will increase coverage by 0.38%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##           master      #71      +/-   ##
==========================================
+ Coverage   80.51%   80.90%   +0.38%     
==========================================
  Files           8        8              
  Lines         544      555      +11     
==========================================
+ Hits          438      449      +11     
  Misses        106      106              

Impacted Files	Coverage Δ
src/instruct.jl	95.91% <100.00%> (+0.24%)	⬆️

---

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 517f68d...4839165. Read the comment docs.

[GiggleLiu]

The register size is too small. Do you have benchmark for the larger size?

[Roger-luo]

20 qubits

Before:

BenchmarkTools.Trial: 
  memory estimate:  688 bytes
  allocs estimate:  17
  --------------
  minimum time:     1.470 ms (0.00% GC)
  median time:      1.501 ms (0.00% GC)
  mean time:        1.500 ms (0.00% GC)
  maximum time:     1.558 ms (0.00% GC)
  --------------
  samples:          279
  evals/sample:     1

after:

BenchmarkTools.Trial: 
  memory estimate:  32 bytes
  allocs estimate:  2
  --------------
  minimum time:     1.379 ms (0.00% GC)
  median time:      1.405 ms (0.00% GC)
  mean time:        1.407 ms (0.00% GC)
  maximum time:     1.489 ms (0.00% GC)
  --------------
  samples:          282
  evals/sample:     1

[GiggleLiu]

But I don't know why do we need new instructs. Why not fixing the U1 instruction directly?
If you think it is necessary, please add some tests.