Gillespie gives the wrong mean when a reaction rate depends explicitly on time.

[luciansmith]

Fixes #1318. The direct method samples the waiting time from a propensity it treats as constant until the next reaction, which is only correct when the rate laws are time-homogeneous. When a rate depends explicitly on time (directly, or through a time-dependent assignment or rate rule), the SSA mean drifts off the ODE: a rate that is zero at t=0 froze the first reporting interval, and coarse output intervals trailed the true mean by about one interval.

Now we check once, on the first step, whether any rate varies with time (the detection adds no random draws, so time-homogeneous models keep the original code path and produce bit-identical results). When a rate does depend on time, the waiting time tau satisfies integral over [t, t+tau] of s(u) du = -log(r1), where s is the total propensity re-evaluated as time advances; we accumulate that integral with an adaptive trapezoidal sub-step and select the reaction using the propensities at the firing time. This is the direct-method form of the integrated-propensity representation in D. F. Anderson, J. Chem. Phys. 127, 214107 (2007). The representation is exact; this implementation realizes it up to the quadrature error of the sub-step, so it is exact in the limit of small steps and converges as the tolerance is tightened, not bit-exact at a finite tolerance.

Adds regression test TimeDependentPropensityMatchesODE, an immigration-death process with a time-dependent immigration rate and a closed-form ODE mean. The time-homogeneous path is byte-for-byte identical to before, and the stochastic test suite is unaffected (its rate laws are all time-homogeneous; the CSymbolTime models use time only in event triggers).

The time-dependence check is a value-based heuristic; inspecting the rate-law expression trees for the time symbol would be more robust, which is noted in the code.