Excellent update! I really like that she's now invisible without the scanner. I wasn't sure about the flashlight at first, but the gameplay definitely benefits from it, even if it's a little cliche (maybe this could be reworked into something more thematically fitting in the future?) 

I do think the battery recharge cooldown is too quick. You can more or less keep everything on all the time as long as you recharge off cooldown, which kinda kills the desperation you'd get from needing to use it sparingly. Also I think stunning her during evac shouldn't teleport her away, unless she's guaranteed to be close to the player's path to the elevator, because right now one flash basically ensures a safe escape.

I'm catching myself here before I spew a bajillion suggestions at you, as I'm sure you've no shortage of ideas and suggestions, as apparent from the comment section (though I'd be more than happy to suggest more if you do want me to). For now I'll eagerly await any future updates you drop, especially for her AI!

I saw in your Dev post you said you didn't want to develop this in a vacuum, but rather use the communities input, so here I (a little part of the community) am, my 2 cents in tow! :)

I'm sure you've heard this plenty, but regardless I'll reiterate, this is a fantastic start! You have a solid gameplay loop, with strong fundamentals and an easily expandable platform. It's a game I actually enjoyed playing, which is not a common occurrence in this genre!

Now I won't pretend to be an expert in game design or anything, but I would say I'm well above average in it. In my opinion, if you truly want this game to flourish, then before you work too much on expanding content and story, I think you need to focus on 2 things.

• The first, and arguably the most important- Her AI

As she is right now, her AI is far too simple. After just a couple runs, I was able to figure out she will 1st) spawn somewhere in the level. 2nd) pathfind straight towards the player. 3rd) only move when the player moves, unless in aggressive mode.
With how loud her footsteps are, it's extremely easy to tell where she's coming from (side note, I believe you should remove these, or at least make them very quiet. I tried playing with the sound off [her footsteps aren't tied to either sound option] and it made the game much more intense). Once you know where she's coming from, all you have to do is watch the door until she walks into your FOV and disappears.  It makes the game trivially easy.

I suggest to improve her AI by having her, much like her source material, stalk the player first. MalO's whole thing is she isn't a physical entity, and is constantly in your peripheral view, driving you crazy. I think you should have her linger around the player for a while before she finally makes her move, building the player's paranoia before she strikes, and keeping them on edge.
This could be achieved by not tying her down to a physical location, allowing her to teleport freely in the facility as long as the player isn't looking there. She quickly pops in the edge of the player's screen, or in a pitch black area, before vanishing to a new location. After she's  stalked the player for a while, she strikes, distorting the screen even heavier than before (or some other subtle tell) where the player will have to drive her off. This brings me to- 

• My second point of focus- How to counter her

Right now, all you do is look at her (or run if she's in aggressive mode, but this is fine). I think this is too easy, and makes dissuading her too trivial. I think there should be some harder way of making her retreat.
This part I don't have as solid of a suggestion for, as it more depends on what direction you want to take the style of the game. Maybe you could have a device that shoos her away but takes battery? Or maybe a quick time event? Or multiple copies of her appear and you need to figure out which is the real one? Or maybe just chuck a gear at her head, there's plenty of options for however you want to take it.
Making her encounter more challenging to avoid adds stress to the player as she makes her approach, causing panic, and possibly mistakes. Adding traps and other challenges that take the players attention in combination with this can overwhelm the player, creating an easily scalable difficulty curve that can be procedurally modified as the game goes on, without feeling confusing or unfair.

But that's just my 2 cents! No matter what, I'm excited to see where you take this game, and wish you the best of luck in all future development!