feat: support release rollback

[riven-blade]

What type of PR is this?

feature

What this PR does / why we need it:

feat: rollback release

Which issue(s) this PR fixes:

Fixes #1028

Special notes for your reviewer:

Does this PR introduce a user-facing change?

feat: support release rollback

Additional documentation e.g., design docs, usage docs, etc.:

feat: support release rollback

[coveralls]

Pull Request Test Coverage Report for Build 12669579227

Details

• 331 of 969 (34.16%) changed or added relevant lines in 19 files are covered.
• 44 unchanged lines in 5 files lost coverage.
• Overall coverage decreased (-1.1%) to 46.415%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
pkg/cmd/cmd.go	0	1	0.0%
pkg/engine/operation/port_forward.go	2	4	50.0%
pkg/engine/apply/utils.go	0	3	0.0%
pkg/engine/apply/state/summary.go	13	17	76.47%
pkg/engine/apply/print_details.go	55	61	90.16%
pkg/cmd/meta/meta_options.go	0	12	0.0%
pkg/cmd/apply/apply_options.go	0	15	0.0%
pkg/engine/apply/port_forward.go	0	17	0.0%
pkg/cmd/preview/preview_options.go	3	24	12.5%
pkg/cmd/apply/apply.go	0	29	0.0%

Files with Coverage Reduction	New Missed Lines	%
pkg/cmd/preview/preview.go	1	31.68%
pkg/engine/operation/port_forward.go	1	32.28%
pkg/engine/resource/graph/storages/local.go	6	7.69%
pkg/util/signal/signal.go	17	0.0%
pkg/cmd/apply/apply.go	19	0.0%

Totals
Change from base Build 12664971436:	-1.1%
Covered Lines:	10823
Relevant Lines:	23318

---

💛 - Coveralls

[liu-hm19]

Thanks for your contribution! We'll review this PR a.s.a.p : )

[liu-hm19]

@riven-blade Hi, I've just reviewed the code you submitted, and I think that it looks good to me overall. However, there are still two points I'd like to confirm:

• For the kusion release rollback command, should it only allow users to roll back to a previously successfully applied release? Since rolling back to a failed revision doesn't seem meaningful, should we add some validation for this in the command?
• Why can't we directly reuse the existing Apply and State related methods? Instead, we have to implement them again, especially creating a new State struct in the release package. Could you please clarify the reasoning behind this?

cc @ffforest @Yangyang96

[riven-blade]

Hi, I have made major adjustments to the abstraction of the previous apply logic, and the different operations between apply and rollback are completed in the prepare stage. At the same time, the place where fixme was identified in the previous apply logic has been fixed. Now there is no problem with the code logic when I test it, but I think there are still areas where the code can be optimized, and the priority is not that high. I don’t want a PR to be so heavy