Allow use of strided vectors with mul! (gemv! and gemm!)

[maleadt]

Fixes #78, fixes #448

Not fully generalized, and simplified some of the signatures, because using the AnyCuArray union everywhere makes package import time regress from ~5 to ~25 seconds.

cc @ChrisRackauckas

[codecov]

Codecov Report

Merging #450 into master will decrease coverage by 0.23%.
The diff coverage is 90.47%.

@@            Coverage Diff             @@
##           master     #450      +/-   ##
==========================================
- Coverage   80.73%   80.50%   -0.24%     
==========================================
  Files         166      166              
  Lines        9085     8833     -252     
==========================================
- Hits         7335     7111     -224     
+ Misses       1750     1722      -28     

Impacted Files	Coverage Δ
lib/cublas/wrappers.jl	92.27% <ø> (ø)
lib/cublas/linalg.jl	94.44% <89.18%> (-1.15%)	⬇️
src/pointer.jl	76.81% <100.00%> (+0.34%)	⬆️
test/cublas.jl	91.17% <100.00%> (+2.36%)	⬆️

---

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 64b9f1d...06466f6. Read the comment docs.