Python3 solution using the built in heapq library:

import heapq

def cookies(k, A):
    heapq.heapify(A)
    
    ops = 0
    while True:
        f = heapq.heappop(A)
                
        if f >= k:
            return ops
        
        if len(A) == 0:
            return -1
        
        s = heapq.heappop(A)
        n = f + 2*s
        heapq.heappush(A, n)
        
        ops += 1

3 days and about 10 attempts (include selfmade binary search for insert value to sorted array). I love it:). I solved this problem using two queues:

In the first queues copy sorted list. In the second queue will write all calculated values (1cookie +2x 2cookie). This queue is automaticaly sorted because values increase . When extracting the value, select the minimum between two queues.

Sorry of bad English, i'm just learning)

https://github.com/smatvienko1978/MyFisrstRepository/blob/ced711c122e5f97b9313e73130b3729dfc1947a8/Program.cs

in the first example they remove the cookie with sweetness '7' twice, once with 6 and once with 8. is this an error or am I misunderstanding the whole remove the cookies part?

Something is up with the performance limits on this one. The number of test cases failing due to timeout ranges from 1 to 3 (literally running the exact same code). One test case in particular I ran in Groovy console and it took less than 0.5 seconds to get the correct answer, which seems like it should be well below the time limit.

Relatively straightforward with the C++ standard library. I used multiset as the primary data structure.