Use taxon children when searching classification

[fkoessler]

Description

In current implementation, eligible? checks the configured taxons and all descendants and actionable? only seems to check against the taxons themselves. This leads to buggy behavior. See: #1409 (comment)

This commit modifies actionable? and have it check against configured taxons and all descendants in order to have a coherent behavior.

[aitbw]

Hey @fkoessler, thanks a lot for submitting this PR! Could you please add some specs to cover this change? 🙏

[fkoessler]

Hello @aitbw , thanks for reviewing. I added some specs that cover the change. Let me know if I can improve the PR.

[aitbw]

Some minor typos to correct and we should be done ☺️

[aitbw]

Thanks @fkoessler! 👌 could you please squash your commits into one? That should be enough to finish this PR 😸

[kennyadsl]

Thanks, this is great! Do you think this can impact performances negatively since we now could potentially have way more database queries?

[fkoessler]

There will be a performance overhead since actionable? now calls to #rule_taxon_ids_with_children. I'm not sure if it is significant or not.

actionable? will be called on each line_item of the order, see Spree::Promotion #line_item_actionable?
The performance overhead will probably depend on the depth of the taxonomy, the number of line_items in the order. Should this be benchmarked?
Performance improvements from #2258 are kept.

In any case, the taxon promotion rule won't work as expected until eligible? and actionable? are made coherent.

Let me know what you think about this.

[kennyadsl]

I'm fine with a performance regression if we are making things work as expected. The overhead will probably be higher depending on how many nested taxons you have and if the promotion is auto-applied (without the promo code), but in that case, I'd left to each store to fix eventual performance issues on their own. Thanks for the accurate explanation.

[kennyadsl]

👍 Thanks!

[jacobherrington]

I am okay with this! Thanks @fkoessler!