Implementation for event driven tink

[gauravgahlot]

Description

This PR provides the implementation for event-driven Tinkerbell base model proposed here.

Note
We would like to keep Watch as the function used to stream events for the new event-driven tink. Therefore, the existing hardware.Watch has been renamed to hardware.DeprecatedWatch.

This will also require some changes in Hegel, so that it can use the hardware.DeprecatedWatch.

Why is this needed

Fixes: #294

How Has This Been Tested?

Unit tests and manual testing.

How are existing users impacted? What migration steps/scripts do we need?

This contains a bc-break. Watch has been renamed to DeprecatedWatch, clients will need to be regenerated and upgraded.

Checklist:

I have:

• updated the documentation and/or roadmap (if required)
• added unit or e2e tests
• provided instructions on how to upgrade

Known issues in progress (fixed)

• an event being received twice on the same stream
• tink hardware watch <id> fails to correctly print the hardware data for new events

[codecov]

Codecov Report

Merging #301 (14badba) into master (1992c1c) will increase coverage by 2.64%.
The diff coverage is 60.95%.

@@            Coverage Diff             @@
##           master     #301      +/-   ##
==========================================
+ Coverage   26.09%   28.73%   +2.64%     
==========================================
  Files          14       19       +5     
  Lines        1280     1385     +105     
==========================================
+ Hits          334      398      +64     
- Misses        921      960      +39     
- Partials       25       27       +2     

Impacted Files	Coverage Δ
grpc-server/events.go	0.00% <0.00%> (ø)
grpc-server/grpc_server.go	0.00% <0.00%> (ø)
grpc-server/hardware.go	1.55% <ø> (ø)
client/informers/informer.go	20.00% <20.00%> (ø)
cmd/tink-cli/cmd/events.go	33.33% <33.33%> (ø)
client/informers/notification.go	88.67% <88.67%> (ø)
client/informers/mock.go	100.00% <100.00%> (ø)
... and 2 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 1992c1c...14badba. Read the comment docs.

[gianarb]

According to the demo @gauravgahlot gave today here some feedback.
I have a couple of things to say about the UX and on the feature itself.

I think we should not use an integer for event types and resource_type we should use strings. They are way more readable as constant and it won't require to actually have a catalog of them. A few bytes to store but who cares. If we start supporting custom events and types it will be impossible to keep a centralized index.

We have to figure out how to effectively deserialize data from bytes to struct and JSON.

The CLI I think should reflect that filters we will have to implement something like --resource-type --resource-id --event-type. The --frame can be a more clear --from 5m (that's should be the default value)

Other than that, great start!

[mmlb]

@splaspood fyi

[mmlb]

we should also ensure the old entries are trimmed away right?

[gauravgahlot]

Updated branch with a rebase, no changes.