Fixes #598

Ref #267

Closes #336

The previous syntax used to call the conditional forecast using the extended path method was problematic. It does not allow to match the endogenous and the shock that are flipped. This is problematic as soon as the forecast periods are heterogeneous or the type of expectation (surprise or perfect foresight) are different. We get rid of the previous syntax.

Added new option for perfect foresight simulations (sim1 routine, available only with stack_solve_algo==0).

Try to reduce the size of the nonlinear system of equations by skipping the (last) periods for wich the residuals are
already (almost) zero. The number of periods is not constant during the Newton, the effective number of periods for
each iteration of the Newton is available in oo_.deterministic_simulation.vperiods.

Closes #498

Closes #112, #514

Closes #507

preprocessor: allow anything of the form dates(*) to be accepted as a valid date by the preprocessor

The preprocessor implementation of "shocks" is broken, and the ref. manual
entries are empty. Nothing is implemented in MATLAB files. It's better to
remove this for now, until a working implementation is done.

Ref. #136

Added an option for setting the number of points where the posterior kernel is evaluated when using mode_check. Reorganized the options_ global structure wrt mode_check.

Also add stub in the doc for the new options

Ref #389

Closes #414

Closes #320

power of negative numbers