feat(insights): Switch to ratios for slow and frozen frames

[markushi]

As described here, we want to move to ratios for slow and frozen frames.

Utilizes division_if to calculate the ratios.

[markushi]

This feels like a hack 🙈, an alternative would be adding them to aggregations, but then we'd need to support this for discover queries as well (see count_if as an example,

sentry/static/app/utils/discover/fields.tsx

Lines 291 to 318 in d10148c

	[AggregationKey.COUNT_IF]: {
	...getDocsAndOutputType(AggregationKey.COUNT_IF),
	parameters: [
	{
	kind: 'column',
	columnTypes: validateDenyListColumns(
	['string', 'duration', 'number'],
	['id', 'issue', 'user.display']
	),
	defaultValue: 'transaction.duration',
	required: true,
	},
	{
	kind: 'dropdown',
	options: CONDITIONS_ARGUMENTS,
	dataType: 'string',
	defaultValue: CONDITIONS_ARGUMENTS[0].value,
	required: true,
	},
	{
	kind: 'value',
	dataType: 'string',
	defaultValue: '300',
	required: true,
	},
	],
	isSortable: true,
	multiPlotType: 'area',

) - not too sure if this would blow stuff up, as the backend support for division_if is tightly bound to releases.

[narsaynorath]

I think there's a cleaner way to get around this, but I have to do a bit of investigation. I'm seeing some odd access paths when it comes to the structure of tableMeta. If you want, we can remove this until I get a chance to poke around a bit more and we can put up a separate PR to handle the sortability of the headers.

The endpoint returns in the meta data that the corresponding type to the division_if statements should be percentages so they should be sortable by default when it gets to this block. I know that the key gets kind of mangled for equations but we should be able to work around that

[narsaynorath]

Just a few comments. I think the conversion for the table looks good but there's probably an alternate way to access the tooltips without editing the type and also fixing the sorting without hardcoding the division_if prefix.

I think we should pull out the field renderer part and have a separate PR. I can take a look deeper tomorrow on how to get around it

[narsaynorath]

I think there's a cleaner way to get around this, but I have to do a bit of investigation. I'm seeing some odd access paths when it comes to the structure of tableMeta. If you want, we can remove this until I get a chance to poke around a bit more and we can put up a separate PR to handle the sortability of the headers.

The endpoint returns in the meta data that the corresponding type to the division_if statements should be percentages so they should be sortable by default when it gets to this block. I know that the key gets kind of mangled for equations but we should be able to work around that

[narsaynorath]

looks good to me! thanks for splitting out the sorting changes we discussed