fix the moving_avg modes

[gabor]

(fixes #98)

when using the moving_avg metric, for certain modes there are coefficients (alpha, beta, gamma) that can be configured. the code used parseInt to parse them, so when floating-point values are entered (for example 0.35), it was parsed incorrectly).

also the MovingAverageEWMAModelSettings type was defined incorrectly.

how to test:

1. make sure you have an opensearch datasource plugin configured ( read https://github.com/grafana/opensearch-datasource/blob/main/devenv/README.md if unsure)
2. open the opensearch datasource in explore
3. you should see a working "count" metric query
4. add another metric, choose moving_avg. configure it:
   • field: choose count
   • in options, choose model=holt-winters
   • set some values, for example:
     • window=5, predict=2, alpha=0.1, beta=0.2, gamma=0.3 (make sure alpha beta and gamma are floating-point values, not integers
5. make sure the browser-devtools are open
6. run the query
7. verify that it did not crash
8. look at the ajax request (_msearch)'s request-payload, it contains two lines of json, the second json should contain the correct alpha,beta and gamma values.
9. change the model to exponentially weighted, and repeat the test
10. change the model to holt linear, and repeat the test
11. go to dashboards, create a panel, and create the above-mentioned query there. verify it works. save the panel, then reload the page, and verify that the query is persisted correctly, meaning alpha, beta and gamma contain the correct values after a page reload. test all 3 models (holt-winters, holt-linear, exponentially-weighted)

[kevinwcyu]

lgtm. saw the values in the json payload as floats.

[kevinwcyu]

Hi @gabor, can this be merged?

[gabor]

hi @kevinwcyu , yes, i think it is ready to be merged.