Mode share sampling for PM and CO2 emissions

[AnnaKS123]

Sampling the mode shares for PM and CO2 emissions uses a Gamma function which creates Dirichlet distributed parameters. This Gamma function does not work properly for small mode shares and low confidence in the mode shares.

In ithim_setup_parameters.R (line 149 for PM and CO2 starts after PM)

if(PM_EMISSION_INVENTORY_CONFIDENCE<1){
total <- sum(unlist(PM_EMISSION_INVENTORY))
parameters$PM_EMISSION_INVENTORY <- list()
for(n in 1:NSAMPLES){
samples <- lapply(PM_EMISSION_INVENTORY,function(x) rgamma(1,shape=x/total*dirichlet_pointiness(PM_EMISSION_INVENTORY_CONFIDENCE),scale=1))
new_total <- sum(unlist(samples))
parameters$PM_EMISSION_INVENTORY[[n]] <- lapply(samples,function(x)x/new_total)
}
}

Gamma function needs updating.