Prediction intervals for naive and mechanistic return rate models

[SOLV-Code]

Naive

have previously flagged this here

current approach:

• if have more than 1 obs: use p10 and p90
• if have 1 obs: multiply mean by 0.5 and 1.5

Return rate

applying same approach for now, but need to discuss if this makes sense for all the variations: wtmean, ,mean, median

[SOLV-Code]

need to think about bootstrap version as per report, but may not need because we have doBoot() fn

Bit more pondering required