Provide more control over the range of the vertical axis in the analytics charts.

Right now, the vertical range is determined by the record with the largest value. This can make the chart difficult to use for analytics that have periodic outliers.

For example, receiving large raids can cause analytics related to viewer counts to be unusually high for a stream. If the streamer's viewer count is otherwise fairly consistent on most streams, the viewer count over multiple "normal" streams might provide more insight than the viewer count on the outlier stream. However, when the vertical axis is expanded to fit the bars for the stream with the large raid, all of the other bars are shrunk to fit the new scale, and subtle slopes over time can get squashed to the point where they are barely visible.

An even more extreme example can be the "follows" analytic after a streamer is hit by a follow bot attack. If a streamer typically gets a single digit number of follows per stream, but on one stream was followed hundreds of times by a bot, that can make the follows chart practically unusable even after the bots accounts are banned, and even if records are grouped by month.

Some suggestions for how this can be improved:

1. Add a field in the chart's interface to set the vertical range of the chart. This could be a free-form number field where the user could type the maximum range, or a pair of fields where the user could type the minimum and maximum values on the vertical axis. Alternatively, it could be a selector where the user could choose a power of 10 to scale the chart to. (10s, 100s, 1000s, etc.)

2. Add the option to view the chart in logarithmic scale. This could help to view "spiky" analytics where outliers are frequent by reducing the height difference between bars of a higher or lower order of magnitude, while keeping all bars visible in the chart. The downside is it could still make subtle slopes difficult to see, and it could be unintuitive to read for people unfamiliar with log-scale graphs.

3. When grouping bars by week or month, provide the option to have the bar show the median value for the analytic in the week or month. This would be a useful statistic to have anyway, and would cause extreme outliers within the groups to be ignored. However, it may be less useful on analytics where the aggregate value is more relevant than the average. (e.g. Revenue per month may be more useful than average revenue per stream during a month.) Being able to show the median per week or per month also will not help if the outlier is the only stream during that time period. That may be remedied by allowing a user to view a moving median over a number of streams rather than a median within a week or month.

4. Provide a way to filter out bars that are "statistically insignificant" based on a standard error calculation. Like with the median, this would be a useful statistic anyway. It would have the advantage of working even when not grouping by week or month, but depending on how the calculation is done, it may or may not be useful when the chart is showing long lengths of time where the streamer has had periods of growth.