haz.to.prob: convert from hazards to probs
In dfeehan/mortfit: Fit and evaluate mortality models

this function takes a hazard function and an age range and returns an approximation of the probability of death implied by the hazard over that age range. it does this either through approximations or through numerical integration. TODO - add more details here NB: this function expects ages to be the starting age (no need to add 0.5 or anything) TODO - document warning for integration try-error TODO - tests here - not too hard to think of...

1	haz.to.prob(haz.fn, theta, ages, nozero = TRUE, approx = FALSE)

`haz.fn`	the hazard function
`theta`	parameter values for the hazard function
`ages`	a vector of ages with entries (z1,z2,...,zk). we'll return probabilities of survival for (z1,z2), (z2,z3), ..., (z(k-1),zk)
`nozero`	if TRUE, replace values that would be returned as zero with very small numbers instead. (this prevents errors when taking logs.)
`approx`	if TRUE, use analytical approximation; if FALSE, use numerical integration