fix memory leak in is_planar

[dcoudert]

Fixes the memory leak in is_planar reported at https://ask.sagemath.org/question/78467/memory-leak-in-is_planar/

We now ensure that memory is released before returning the answer.

📝 Checklist

• The title is concise and informative.
• The description explains in detail what this PR is about.
• I have linked a relevant issue or discussion.
• I have created tests covering the changes.
• I have updated the documentation and checked the documentation preview.

⌛ Dependencies

[dcoudert]

@videlec I'm not sure how to show that this fixes the memory leak. At least when I check on my laptop the memory usage with htop, it's not increasing anymore when running this very long example

sage: count = 0
....: for g in graphs.nauty_geng('-C 10'):
....:     if g.is_planar():
....:         count = count+1
....: print(count)
545808

This is consistent with https://oeis.org/A021103

[github-actions]

Documentation preview for this PR (built with commit 669f0d5; changes) is ready! 🎉
This preview will update shortly after each push to this PR.

[user202729]

@dcoudert Maybe you can look into tests/memcheck? #36046 is an example pull request that adds a test for memcheck.

[dcoudert]

I found a way to check the memory consumption. However I'm not sure I can use that to make a doctest. At least we can see here that the memory leak is fixed with this PR.

import psutil
from time import sleep

def get_process_memory():
    process = psutil.Process(os.getpid())
    mem_info = process.memory_info()
    return mem_info.rss

def my_count(n):
    leak = get_process_memory()
    count = 0
    for g in graphs.nauty_geng(f'-C {n}'):
        if g.is_planar():
            count += 1
    sleep(2)  # to ensure that subprocesses have released memory 
    leak = get_process_memory() - leak
    return count, leak

Before:

sage: my_count(8)
(2893, 11714560)
sage: my_count(9)
(36496, 350650368)

With this PR:

sage: my_count(8)
(2893, 0)
sage: my_count(9)
(36496, 0)
sage: my_count(10)
(545808, 0)

[tscrim]

There generally are not good ways to check against memleaks using doctests, even with #36046 IMO. Nor do I think we should add (doc)tests each time we encounter and fix a memleak, at least when it is straightforward. As such, I am setting this to a positive review.

[dcoudert]

Thanks you.