Prospective functions contradict Watanabe et al results?

Greetings!

I've been experimenting with this library, and so far it seems very useful! However, I am confused by the results of `pGWP` functionality.

In Watanabe's results pGWP for CH4 is often higher than AR6 values, particularly for lower RCP scenarios.
Your results do not seem to reflect this, though.

## Example

Based on the binder code, I have:

assuming a dynamic inventory with 1 kg of CO2 and 1 kg of CH4

```python
prospective_functions = {
FLOW_IDS["CO2"]: prospective_characterize_co2,
FLOW_IDS["CH4"]: prospective_characterize_ch4,
}

prospective.set_scenario(iam="MESSAGE", ssp="SSP2", rcp="2.6")

characterize(
inventory_supported[["date", "amount", "flow", "activity"]],
metric="pGWP",
characterization_functions=prospective_functions,
fixed_time_horizon=False,
time_varying_re=False,
fallback_to_ipcc=False
)
```

Which gives:

CO2 seems correct, but CH4 is far below the expected value. No prospective case in Watanabe's study gives values below IPCC AR6 estimates (28).

Start by reproducing the Binder example with the MESSAGE SSP2 RCP 2.6 scenario and the pGWP settings shown. Trace prospective_characterize_co2 and prospective_characterize_ch4, then compare the CH4 result with Watanabe et al. and the IPCC AR6 value. Done means identifying whether the discrepancy is expected or correcting the prospective CH4 characterization and documenting the result.

Written by the indexing model from the issue text.