Fix acceleration/cron recipe: version floor v1.0 → v1.4.0 and stale accelerated size - #598

Summary

• Version floor. The recipe says Works with v1.0+, but its whole subject — the refresh_cron acceleration parameter added in Step 2 — did not exist until v1.4.0. A reader on v1.0–v1.3 cannot complete the recipe. Same correction already merged for the views recipe in Fix views recipe: correct version floor to v1.4.0+ #538.
• Stale accelerated size. Step 2's transcript reports 399.41 MiB for taxi_trips; Step 1 of the same recipe reports 399.38 MiB, which is what the runtime prints today. A README that contradicts itself about the same dataset is drift regardless of which run is newer.

Verified against

spiceai/spiceai — refresh_cron first appears at v1.4.0:

$ for t in v1.0.0 v1.1.0 v1.2.0 v1.3.0 v1.4.0; do
    printf "%-8s %s\n" $t "$(git grep -c refresh_cron $t -- crates | wc -l)"; done
v1.0.0   0
v1.1.0   0
v1.2.0   0
v1.3.0   0
v1.4.0   6

Current definition: crates/spicepod/src/acceleration/mod.rs:502, consumed at crates/runtime/src/init/scheduler.rs:219-254.

Evidence

Ran the recipe end-to-end on CLI/runtime v2.1.5 (spice add spiceai/quickstart, added refresh_cron: "*/30 * * * * *" to spicepods/spiceai/quickstart/spicepod.yaml, spice run):

2026-08-19T12:20:12.712184Z  INFO runtime::init::dataset: Dataset taxi_trips registered (s3://spiceai-demo-datasets/taxi_trips/2024/), acceleration (arrow), results cache enabled. duration_ms=0
2026-08-19T12:20:12.713847Z  INFO runtime::accelerated_table::refresh_task: Loading data for dataset taxi_trips
2026-08-19T12:20:15.315898Z  INFO runtime::accelerated_table::refresh_task: Loaded 2,964,624 rows (399.38 MiB) for dataset taxi_trips in 2s 602ms.
2026-08-19T12:20:30.320838Z  INFO runtime::accelerated_table::refresh_task: Loading data for dataset taxi_trips
2026-08-19T12:20:37.331335Z  INFO runtime::accelerated_table::refresh_task: Loaded 2,964,624 rows (399.38 MiB) for dataset taxi_trips in 7s 10ms.
2026-08-19T12:21:00.335406Z  INFO runtime::accelerated_table::refresh_task: Loading data for dataset taxi_trips

The cron schedule fires on the 30-second boundary as documented, the spicepods/spiceai/quickstart/spicepod.yaml path is correct, and the six-field expression parses (crates/scheduler/src/channel/cron.rs:39-44 builds the parser with Seconds::Optional). Row count and size confirmed as 2,964,624 / 399.38 MiB.

Load durations in the transcript are left as-is — they vary per run and per network. The Initializing dataset taxi_trips line in the same block is corrected separately by #587; this PR deliberately does not touch that line.