A series of changes to improve the reliability and performance of tictac aae full-sync, and further protect a cluster should issues with full-sync occur (that might otherwise lead to excessive resource consumption).

The primary changes involved are:

- The ability to use day, hour and range -specific checks, not just all checks when making nval-based comparisons between clusters.

- Backoff scheduled work aggressively if work starts to fall behind.

- Improved logging during checks to give operators a view of deltas being discovered by full-sync, to better inform operator interventions.

- Improvements in underlying repos to accelerate cluster-wide operations constrained by last-modified dates.

- The adding of a last-modified date to Riak tombstones so that they can be correctly identified by that date during full-sync checks

- Improvements to underlying backend performance and default configurations, that had been included in the 2.9.8 release.

- A "getting-started" guide to help with the initial configuration of full-sync, and to highlight some key operational tips.

- Simplification, and some validation of riak_client functions to make them easier to use by operators.

- More run-time configuration overrides to provide a further safety net should issues arise in production.