0.9-1 First release

0.9-3

- fix problem with exact2x2CI, gave wrong confidence interval for some 2x2 tables
- fix conf.int=FALSE to work correctly for 'blaker' and 'minlike' methods, still ignored for 'central'
- fix p-value for blaker test to be always less than or equal to 1
- fix blaker.exact and fisher.exact to work correctly when y!=NULL
- minor editing of some man pages
- minor editing of some code for clarity