Fix insert_vinsts and use its recursive argument

[abdelq]

Summary

The intersection is done at the wrong spot leading to empty set. Also recursive was not being used.

Test Plan

Tested on private repo's code.

[abdelq]

Maybe write and write_bytes should also drop the insert_into loop if it's handled here?

[dansgithubuser]

I'm curious what's the point of the & self.kcl.tkcells.keys(). Is it ever not a no-op?

[sebastian-goeldi]

I'm curious what's the point of the & self.kcl.tkcells.keys(). Is it ever not a no-op?

In case it's been manually manipulated yes. This can theoretically happen if someone changes kcl manually. It can crash really hard if that happens, and the klayout errors can be very no fun when that happens

Sorry, maybe I misremember. But I believe if you clean up the layout object, it can have an impact, but I can't remember exactly why I added it. But it was related to having klayout segfault bc accessing an invalid index at some point

[sebastian-goeldi]

Maybe write and write_bytes should also drop the insert_into loop if it's handled here?

yes that sounds good, can you add it to the PR?