cli: execute benchmark plan and implement result evaluation and verdict logic for bisect run

This patch includes the implementation of the `bisect run` subcommand
in the CLI. It resolves and updates the bisection template by injecting
kernel image and module paths, then executes the benchmark plan using
plan_exec to generate new result data.

After execution, the results are loaded and joined with metadata tables
to build a comprehensive dataset. The data is filtered by benchmark and
resultclass, with support for partial string matches and normalized
profiles. The latest run is assumed as the test profile.

It uses show.pivot_results() to compute statistical summaries, and
reduces the pivoted output using slicing to match the expectations of
compute_change(). The change status is extracted and reported as
regression, improvement, or no significant change.

This enables automated result analysis and verdict detection as part of
kernel regression bisection.

Signed-off-by: Aishwarya TCV <aishwarya.tcv@arm.com>