feat: grafana add p99, apiserver, tablet query

[vagetablechicken]

• p99, apiserver, tablet query prc metrics
• overview expand, others collapse
• delete "deploy query count total"
• multi-value for variable "db", "deployment"

[github-actions]

Monitoring integration test

6 tests   - 1   6 ✔️  - 1   1m 1s ⏱️ - 1m 0s
1 suites ±0   0 💤 ±0 
1 files   ±0   0 ❌ ±0 

Results for commit 4cd9a6b. ± Comparison against base commit e229b16.

This pull request removes 1 test.

tests.test_exporter ‑ test_deploy_response_time

♻️ This comment has been updated with latest results.

[aceforeverd]

make all rows (except the first row 'OpenMLDB Cluster Overview') in dashboard default closed, like original

[aceforeverd]

May have more detailed name for the two row?

Separate 'Tablet Detail' in different panel for different tablet endpoint may better. Also API server may have multiple instance

[aceforeverd]

db & deployment variable should support multi-value

[aceforeverd]

Remove 'Deploy Query Count Total' in OpenMLDB Cluster Overview

[vagetablechicken]

May have more detailed name for the two row?

Separate 'Tablet Detail' in different panel for different tablet endpoint may better. Also API server may have multiple instance

Each apiserver will be an instance, no need to seprate
And I want Tablet Detail to be the one, so we could see the difference in one panel, don't need to check multi panels to know some tablet is slow or heavy

[aceforeverd]

May have more detailed name for the two row?
Separate 'Tablet Detail' in different panel for different tablet endpoint may better. Also API server may have multiple instance

Each apiserver will be an instance, no need to seprate And I want Tablet Detail to be the one, so we could see the difference in one panel, don't need to check multi panels to know some tablet is slow or heavy

I still think there is too many series on single graph. To compare perf among tablets, then separate with different tp value ?