Scores differing from manual lighthouse tests

[LunarDevelopment]

Hello,

I'm getting consistent scores on all metrics except "Performance" which is sometimes differing wildly from a manual lighthouse scan performed in Chrome directly. e.g. scoring 98 from this library vs 30 with chrome lighthouse scan.

This affects both the Mobile and Desktop form factors.

Any idea why this is ?

v 1.2.2 of the library used.