Support $\Delta_t$ ramps

[pbauman]

I won't have time to do this for a couple of weeks, but wanted to capture the thought and solicit comments.

For these laminar flame problems, you really need to time step to steady state on coarse grid (barring having an initial guess based on a flame sheet model (e.g. ala Smooke, Mitchell, Keyes, 1989) or something similar). During the initial solution phase, time steps are a little more restricted while the chemistry starts up (flame ignites), but we can take much larger steps later as the chemistry settles in. Would be nice to build in the time step ramp in the input file. Was thinking something like:

delta_t = '20:1.0e-5 100:1.0e-3 100:1.0e-2'

This would indicate three stages of time-stepping: first, take 20 steps with $\Delta_t = 1.0e-5$ followed by 100 steps of $Delta_t = 1.0e-3$ finally followed by 100 steps of $\Delta_t = 1.0e-2$.

The : serves as the separating symbol for parsing. We'd still be backward compatible with a single $\Delta_t$ --- we just check the size of delta_t: if it's 1, we do what we've always done, if > 1 then we prepare this time-stepping.

Thoughts? Better syntax ideas?