Fix truth vector generation in forecastbaselines vignette

The code was trying to access out-of-bounds indices (trend[51:62] when
trend only has 50 elements), which created NA values that became Missing
in Julia, causing a type error.

Fixed by generating future trend and seasonal values for the forecast
horizon instead of trying to index beyond the data length.

Error was: "expected Vector{Float64}, got Union{Missing, Float64}"

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>

Author: sbfnk

vignettes/forecastbaselines.Rmd

@@ -218,7 +218,10 @@ To evaluate forecast accuracy, you need to provide the true values.
 ```{r scoring}
 # Generate some "future" observations
 set.seed(456)
-truth <- trend[n + 1:12] + seasonal[n + 1:12] + rnorm(12, sd = 2)
+future_time <- (n + 1):(n + 12)
+future_trend <- 0.5 * future_time
+future_seasonal <- 5 * sin(2 * pi * future_time / 12)
+truth <- future_trend + future_seasonal + rnorm(12, sd = 2)
 
 # Add truth to forecast
 fc_with_truth <- forecast(