Improve power report calculation performance by omitting FullDPS and eHP unless needed

[AdamZ-8113]

Description of the problem being solved:

The power report feature currently runs with this logic:

if (useFullDPS ~= false or build.viewMode == "TREE") and usedFullDPS then
    calcs.calcFullDPS(...)
end

Because we're always using the build.viewMode == "TREE", we're always using "Full DPS", even when the power report selected in PoB is something that doesn't require it, like "Hit DPS" or "Life".

Each report also calls calcs.perform(env) without skipEHP, so eHP is also being calculated every time. So basically if you want "Hit DPS" you're getting Hit DPS + Full DPS + eHP even though the power report only renders Hit DPS.

This change fixes this issue by only calculating Full DPS and eHP when the selected power report actually requires them.

In a benchmark test I ran with 29 builds across 10 different power report metrics, results were overwhelmingly positive. Histogram of the performance improvements below:

Steps taken to verify a working solution:

• Built a benchmark tool that used "all depth" and ran benchmark across 10 different power report metrics - Histogram of the data provided above.
• Manually compared and verified different power reports across many different builds
• Checked both speed and correctness

Link to a build that showcases this PR:

Import this build into the current dev branch without the PR applied and run a Power Report for anything that doesn't require FullDPS or eHP, such as Attack/Cast Speed, Average Hit, Hit DPS, etc. Then apply the PR and compare the speed of calculation. It's significant enough that you can count the full seconds of timing difference. This is because "Full DPS" is very slow, and by skipping Full DPS when it's not necessary, the Power Report is much faster.

https://pob.codes/b/-MrLSBaNi7Q

Before screenshot:

N/A

After screenshot:

N/A