Before
---

- When opening the appointment page, the possible capacities that a user can book needs to be calculated to show the capacity dropdown.
- This is currently done by getting all the possible combinations that could arise from each resource along with its linked resources (Linked resources are resources that are used to combine with the main resource to allow a bigger capacity)
- The complexity of this approach blows up with the increase in linked resources. For each resource we end up with O(2^m) where m is the number of linked resources it has. If they have the same number of linked resources, and n is the number of resources, we end up with O(n*2^m)

Solution:
- Flipping the algorithm, we create a structure for possible capacities and only keep the best combination for a found capacity.
- We first store all available capacity for relevant resources.

Algorithm
---

For a resource:
- We initialize the solution S to {0: []}
- Greedy dynamic programming is then used to get all the possible combinations by building the result dict, adding one resource at a time. We add its capacity to all entries of S -> we use the combination if a new capacity is reached OR update the combination if the number of elements of combination is less than the existing one.

Then, for the global solution:
- We go through S of resources (in the order of self), and add entries to G (the general dict of solutions). Again, on collision, we only keep the lower-cardinal combination.

The complexity of this algorithm is O(n*u) where u is the dynamic programming complexity. u would be quadratic O(m^2) if the sub-sums of capacities overlap heavily, for instance when resources have the same capacity, like tables of a restaurant, but could reach (2^m) in the worst case scenario.

Related changes / side effects
---

1. Fixing an issue

2. Cleaning strange logic

3. Cleaning max capacity computation

Benchmark:
- opw-5177932 goes down from 6 mins to less than 2 seconds using the new algorithm.

Also include
---

Fix the max capacity on resources when skipping resource selection. It computed the max capa based on values that were not yet emptied in the website_appointment controller.


Task-6233563

Forward-Port-Of: odoo/enterprise#118981