standardize wlr and maxcombo output

[LittleBeannie]

This PR covers:

1. Unify the output of wlr and maxcombo as a list containing

   • method,
   • parameter,
   • estimate,
   • se,
   • z
   • p_value and corr (only for maxcombo).

2. Cleanup fh_weight, mb_weight and early_zero_weight, which only assigns weights to subjects. And the calculation of the z-score, estimation, se, is done in wlr().

3. Update maxcombo.

# Old workflow
sim_pw_surv(n = 200) |>
   cut_data_by_event(150) |>
   counting_process(arm = "experimental") |>
   fh_weight(rho_gamma = data.frame(rho = (0, 0, 1, 1), gamma = (0, 1, 0, 1)), return_corr = TRUE) |>
   pvalue_maxcombo()

# New workflow
sim_pw_surv(n = 200) |>
   cut_data_by_event(150) |>
   maxcombo(rho = c(0, 0), gamma = c(0, 0.5), return_corr = TRUE)

5. Revise sim_fixed_n to fit into the new output of wlr or maxcombo.
6. Update related tests.