Luca,
I see three bends. (I think. Or is the spring tangled in the last couple coils. If so there is an easy way to undo the tangle.)
Assuming they are all bends, start with the two on the outer coil. The way to deal with them is to "circle the regulator". You first need to level that upward bend. Couple ways to do that, maybe easiest on a glass plate. Then install the spring only back into the balance bridge. Turn it upside down. Now use the hole jewel to center the collet. Next use the regulator to form the the section of the spring that acts between the pins.
The other bend in the third or 4th coil would deem this spring unuseable for precision timing. It was proven to me in Switzerland that you can only correct deformed springs in the first coil from the pinning point or the stud and still get precision timing results.
Yes I know all the stories about "a guy" who perfectly formed a coil after he had pulled it straight and such. All myths. Ask someone to show you and to prove it on the timing machine across the 8 vertical positions. Ain't gonna happen. If Tony Simonin said he cannot do it, then that is it.
If it IS tangled, then there is the only useful thing I ever found in Fried's books. Insert a piece of paper (I keep some 35mm film for this and other things like a dial protector when using levers) into the balance spring past the tangle. Now gently rotate the spring so that the tangle is moved to the stud. You may have to lift the second coil over the stud to clear the tangle, but you are there.
If you cannot get a replacement spring from a donor movt, then you have to do the best you can. Use a microscope and two needles to nudge the spring as near to the bend of that coil as you can. The more moves you attempt, the worse trouble you will cause.
I use a designated broach fitted into a handle for holding the collet for rough corrections. I turn broach with the mounted balance spring broach point resting on the bench. Then I mount the spring on the staff and use calipers. I first level by checking its appearance as it spins. Then I look from the top and check to ensure it looks like an old vinyl record with it flowing out from the center to the stud. You need to thin the arms of the caliper to be able to see the sring well from the top. Levin did make a special caliper for this.
I use needles and reworked stick oilers for everything but that upward bend in your outer coil. That requires two good tweezers, one to hold the spring and the other to rotate to correct the upward bend. This is the only time I use steel tweezers BTW. Bronze are nicer to the parts and they are easy to dress with a file. I bought a dozen pair of 3c bronze in 2010 and I am still working with the first two.
I just corrected a small Longines cocktail watch bought back from WWII by a pilot interned in Switzerland. Used a very soft spring (early alloy that is like the stuff used in the Font 69). I see you also have an alloy spring with mono balance. It was grazing the balance bridge and the previous worker could not see it due to size and clearance, so he added two timing washers to slow the rate by 6 minutes. This watch was only made between 1941 and 1944.
I HAD to use a microscope at 30X and quite honestly, I shudder to think what I pretended I could see 30 years ago. But, I was able to correct the leveling because rather than having my nose in the way with a 10X loupe I had 6 inches of working distance with the benefit of 30 x mag. Removed the washers and it holds a respectable rate across 3 positions (30 seconds variance) over 24 hours with the regulator centered.