propagate inbounds

[Roger-luo]

not a big problem, since julia know how to hoist the boundscheck sometimes now, but some of the instruct didn't correctly inbounds the callee

before:

julia> @benchmark instruct!(st, Val(:T), (3, )) setup=(st=rand(ComplexF64, 1<<15))
BenchmarkTools.Trial:
  memory estimate:  0 bytes
  allocs estimate:  0
  --------------
  minimum time:     17.568 μs (0.00% GC)
  median time:      19.893 μs (0.00% GC)
  mean time:        20.972 μs (0.00% GC)
  maximum time:     130.544 μs (0.00% GC)
  --------------
  samples:          10000
  evals/sample:     1

after

julia> @benchmark instruct!(st, Val(:T), (3, )) setup=(st=rand(ComplexF64, 1<<15))
BenchmarkTools.Trial:
  memory estimate:  0 bytes
  allocs estimate:  0
  --------------
  minimum time:     14.798 μs (0.00% GC)
  median time:      17.689 μs (0.00% GC)
  mean time:        18.589 μs (0.00% GC)
  maximum time:     109.593 μs (0.00% GC)
  --------------
  samples:          10000
  evals/sample:     1

[codecov]

Codecov Report

Merging #34 into master will increase coverage by 0.07%.
The diff coverage is 92.1%.

@@            Coverage Diff             @@
##           master      #34      +/-   ##
==========================================
+ Coverage    80.6%   80.68%   +0.07%     
==========================================
  Files           8        8              
  Lines         495      497       +2     
==========================================
+ Hits          399      401       +2     
  Misses         96       96

Impacted Files	Coverage Δ
src/utils.jl	83.69% <92%> (+0.36%)	⬆️
src/instruct.jl	95.38% <92.3%> (ø)	⬆️

---

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 6ee8ba1...097ab5c. Read the comment docs.