…nfiguration management (#17783)

| Q               | A
|-----------------|-----
| Branch?         | 2.0 <!-- see the comment below -->
| Bug fix?        | no
| New feature?    | no
| BC breaks?      | no
| Deprecations? | no <!-- don't forget to update the UPGRADE-*.md file
-->
| License         | MIT

Continuation of #17782

This proposal is motivated by the issue in Behat when suite
configuration includes contexts that define the same method.
In brief, this decoupling provides more flexibility in managing suite
configurations.

…en I add` (#17785)

| Q               | A
|-----------------|-----
| Branch?         | 2.0 <!-- see the comment below -->
| Bug fix?        | no
| New feature?    | no
| BC breaks?      | no
| Deprecations? | no <!-- don't forget to update the UPGRADE-*.md file
-->
| License         | MIT

Continuation of:
- #17782
- #17783

Until now, the checkout-related scenarios have been set up as follows:
• UI – via web actions (e.g. `Given I have a product ...` actually
triggered browser actions to prepare the initial step),
	•	API – a mix of messenger commands and API calls.

During the development of Sylius 2.0, with the introduction of dynamic
components, all scenarios involving product-related actions had to be
marked as JS to work correctly. This significantly increased build time
and reduced overall stability.

This PR focuses on decoupling the Behat architecture to unify the test
setup for both UI and API. The goal is to eliminate most JS tags, which
should lead to greater stability. Some performance improvements are
already noticeable.


This PR doesn’t yet include all refactored scenarios, but at this stage
we already observe the following performance gains:
• For non-JS scenarios, the timing is likely similar — some JS scenarios
have been converted to non-JS, but at the same time, we benefit from
faster setup thanks to the refactor (using messenger commands instead of
API calls or UI interactions to prepare tests).
• For JS scenarios executed via Chromedriver, the same applies — some
Panther scenarios have been switched to Chromedriver, and performance
remains comparable.

JS with Panther: **~30 min → ~20 min**

…en I add` - Promotion & Some other scenarios (#17818)

| Q               | A
|-----------------|-----
| Branch?         | 2.0 <!-- see the comment below -->
| Bug fix?        | no
| New feature?    | no
| BC breaks?      | no
| Deprecations? | no <!-- don't forget to update the UPGRADE-*.md file
-->
| License         | MIT

Continuation of:
- #17785
- #17782
- #17783

Until now, the checkout-related scenarios have been set up as follows:
• UI – via web actions (e.g. `Given I have a product ...` actually
triggered browser actions to prepare the initial step),
	•	API – a mix of messenger commands and API calls.

During the development of Sylius 2.0, with the introduction of dynamic
components, all scenarios involving product-related actions had to be
marked as JS to work correctly. This significantly increased build time
and reduced overall stability.

This PR focuses on decoupling the Behat architecture to unify the test
setup for both UI and API. The goal is to eliminate most JS tags, which
should lead to greater stability. Some performance improvements are
already noticeable.


This PR doesn’t yet include all refactored scenarios, but at this stage
we already observe the following performance gains:
• For non-JS scenarios, the timing is likely similar — some JS scenarios
have been converted to non-JS, but at the same time, we benefit from
faster setup thanks to the refactor (using messenger commands instead of
API calls or UI interactions to prepare tests).
• For JS scenarios executed via Chromedriver, the same applies — some
Panther scenarios have been switched to Chromedriver, and performance
remains comparable.

JS with Panther: **~30 min → ~20 min**

…` and related scenarios) (#17816)

| Q               | A
|-----------------|-----
| Branch?         | 2.0 <!-- see the comment below -->
| Bug fix?        | no
| New feature?    | no
| BC breaks?      | no
| Deprecations? | no <!-- don't forget to update the UPGRADE-*.md file
-->
| License         | MIT

Continuation of:
- #17785
- #17782
- #17783

Until now, the checkout-related scenarios have been set up as follows:
• UI – via web actions (e.g. `Given I have a product ...` actually
triggered browser actions to prepare the initial step),
	•	API – a mix of messenger commands and API calls.

During the development of Sylius 2.0, with the introduction of dynamic
components, all scenarios involving product-related actions had to be
marked as JS to work correctly. This significantly increased build time
and reduced overall stability.

This PR focuses on decoupling the Behat architecture to unify the test
setup for both UI and API. The goal is to eliminate most JS tags, which
should lead to greater stability. Some performance improvements are
already noticeable.


This PR doesn’t yet include all refactored scenarios, but at this stage
we already observe the following performance gains:
• For non-JS scenarios, the timing is likely similar — some JS scenarios
have been converted to non-JS, but at the same time, we benefit from
faster setup thanks to the refactor (using messenger commands instead of
API calls or UI interactions to prepare tests).
• For JS scenarios executed via Chromedriver, the same applies — some
Panther scenarios have been switched to Chromedriver, and performance
remains comparable.

JS with Panther: **~30 min → ~20 min**