refactor(ecstore): use shared ForegroundPressure for data movement
The data movement backpressure module carried its own byte-identical copy of ForegroundPressure, its reason() label mapping, and the foreground utilization computation. rustfs-concurrency now owns that logic as workload::ForegroundPressure and workload::foreground_pressure, so the local copy was a cross-crate synchronization point that could silently drift from the heal-side and admission-side behavior.
Delete the local type and computation and call the shared function instead. The call site keeps what is specific to data movement: the config.enabled short circuit, the optional provider unwrap, and the read/write threshold percentages read from DataMovementBackpressureConfig. The reason() labels emitted into the rustfs_data_movement_backpressure_total metric and the data_movement_backpressure log event are unchanged, as are the existing tests and their assertions.
Refs rustfs/backlog#2048
(cherry picked from commit 6a26e144e06ced53a8dfd1712ab7aa24589646ff)
(cherry picked from commit ab5ab417e80179265c32b22a5e671eac0b9e43ae)
