Trapezoidal integration of stellar age weights

[aphearin]

Trapezoidal integration of stellar age weights

[aphearin]

This PR is motivated by the observation that for some edge cases where the SFH table of a galaxy is tabulated very near the edge of the stellar age bins used to calculate the PDF of stellar ages, DSPS appears to have numerical noise in estimating P(τ) for very young stars. In experimenting with particularly problematic examples, trapezoidal integration for the calculation of the log10(SMH) table appears to resolve the issue.

Note that the stellar age PDF calculation still uses the _jax_get_dt_array function to define the τ_age bin edges from the SSP τ_age grid. It's only in the calculation of cumulative Mstar at the bin edges that _jax_get_dt_array is no longer used. This plot shows an outlier example where the effect is extreme.