Merge branch 'master' of https://github.com/robjhyndman/forecast

* 'master' of https://github.com/robjhyndman/forecast:
  fixed a bug about the wrong frequency in hw function
  changed the position of the sub-functions in HoltWintersNew.R
  add HoltWintersNew.Rd, modified NAMESPACE, tried to compile the whole package, but failed with error: 'collect2: ld return 1',  it seems related to the compilation environment.
  add HoltWintersNew.Rd, modified NAMESPACE, tried to compile the whole package, but failed with error: 'collect2: ld return 1',  it seems related to the compilation environment.
  fixed bug about the 'parameter out of range' caused by the checkpara() function
  simply test the multiplicative HoltWintersNew(), all ok for different smoothing parameters. Please see new bug_fix_list_Zhenyu.txt. Changed some small errors in HoltWintersNew.R
  add multiplicative, have not systematically tested yet
  all bugs fixed, see detail in bug_fix_list_Zhenyu.txt, the next step is considering multiplicative
  new HoltWintersNew.R file with changes listed in the bug_fix_list_Zhenyu.txt git status git status  cancel
  add bug_fix_list_Zhenyu.txt file to list all the test result and bugs
  test, the same to previous one
  test... the same to the previous commit
  changed HoltWintersNew. R, test it for different initial smoothing parameters. It has bug while using:  HoltWintersNew(x, gamma=FALSE). I am fixing the bug.
  added ./R/HoltWintersNew.R, basic functions with filter and parameter initialisaton. I only consider for case with all level, trend and seasonal, have not consider the case without trend or seasonal, have not considered the mutiplicateive case, still need polish and test. You could test it first and give me some suggestion if you have time. Thanks