Merge pull request #18 from hjmshi/lbfgs-errors

clean up code and correct computation of gtd

remove tensor initialization for lr quantities

include line search debug mode, allow for different dtypes, and small bug fixes

fix small bug in gp regression example

add license

Update README.md

small change to full batch example

fix small bug in step function

add gpytorch example

add full-batch L-BFGS wrapper

correct L-BFGS inplace and update README and CUTEst tests

update README.md

update README.md

add inplace option