Tweak best practices for LCJS bench

[niiloArction]

1. Disable theme effects for performance benchmarks. This is enabled by default, but does not reflect best performance.
2. Fix incorrect syntax for line series data pattern.

I didn't find instructions on how the benchmark results were measured, so not sure how this would reflect on them. But with requestAnimationFrame load speed test the loading seems to be ~100 milliseconds faster.

We were a bit late on updating our own performance test bench code with the latest best practices, so its understandable that some of this was missed. Would appreciate updating the results but I totally understand if its not done immediately.

[leeoniya]

tested with your changes. the scripting time improvement is ~30ms (out of 250ms).

i normally do forced GC between continuous page reloads to purge JIT cache, to force re-parsing of libs, and simulate fresh boot. this affects large libraries much more than small ones. without forced GC, i get 130ms scripting for LightningChart but with forced GC i get 240ms scripting (with Evaluate Script taking ~85ms).

for reference, uPlot is ~40ms for both GC and non-GC runs, with Evaluate Script < 2ms.

i'll consider re-doing the bench without forced GC in the future, but i also i want to reward libs for striving for optimized bundle size, so it's not clear that this specific metric should be changed to omit this cost.

uPlot with GC:

uplot-gc.mp4

---

uPlot no GC:

uplot.mp4

---

LightningChart with GC:

lightning-gc.mp4

---

LightningChart no GC:

lightning.mp4