`sim.fmsm` gives different number of transitions when simulating same sample size

[mafed]

Simulating the same number of subjects from the bosms3 dataset doesn't give the same number of transitions. I wasn't expecting exactly the same numbers but these differences seem weird. Here below you have an MRE: I used both tidy_sim.fmsm() (defined in #163) and just looked at number of first transitions. Might this be related to #153?

### MRE sim.fmsm 
library(tidyverse)
library(flexsurv)

blnorm <- flexsurvreg(Surv(years, status) ~ trans, data = bosms3, 
                      dist = "lnorm")
tmat <- rbind(c(NA,1,2),c(NA,NA,3),c(NA,NA,NA))
set.seed(777)
invisible(rnorm(1e3))
sim_blnorm <- sim.fmsm(bexp, M = nrow(bosms3), t = max(bosms3$years), 
                       trans = tmat) 
sim_blnorm_tidy <- 
  sim_blnorm %>% 
  tidy_sim.fmsm()

bosms3 %>% count(status, trans)
# An object of class 'msdata'
# 
# Data:
#   status trans   n
# 1      0     1 *101*
# 2      0     2 175
# 3      0     3  35
# 4      1     1 *103*
# 5      1     2  29
# 6      1     3  68

sim_blnorm_tidy %>% count(status, trans_num)
# # A tibble: 6 × 3
#   status trans_num     n
#    <int>     <dbl> <int>
# 1      0         1    *56*
# 2      0         2    56
# 3      0         3    81
# 4      1         1   *345*
# 5      1         2   110
# 6      1         3   264

sim_blnorm$st %>% as_tibble() %>% count(V1 == 1, V2 == 2)
# # A tibble: 2 × 3
#   `V1 == 1` `V2 == 2`     n
#   <lgl>     <lgl>     <int>
# 1 TRUE      FALSE       166
# 2 TRUE      TRUE        *345*