feat(hub): Added support for multiple root ports

[tore-espressif]

This will allow having both FS and HS root ports active at the same time and connected devices working properly.

---

Note

Medium Risk
Touches core USB host hub/enumeration plumbing and root-port state management, which can affect device attach/detach, enumeration ordering, and suspend/resume behavior across ports. Changes are scoped but concurrency/ISR-driven paths and multi-port edge cases increase regression risk.

Overview
Adds multi-root-port support across the USB host stack. The Hub driver now initializes/powers multiple HCD root ports based on hub_config->port_map, tracks per-port state/requests, and routes ISR events and recovery/disable actions per root port.

Makes devices root-port aware end-to-end. Device tree nodes and usbh_devs_add() now associate devices with a specific root_port_hdl, adds usbh_dev_get_root_port_hdl() and ext_hub_get_root_port(), and tightens the “address 0” constraint to only conflict within the same root port.

Improves enumeration scheduling for multi-port scenarios. Enumeration starts by UID via new usbh_devs_open_uid(), and adds a small pending-UID queue so concurrent attach events enqueue and enumerate sequentially (returning ESP_ERR_NO_MEM if the queue is full).

Test/build updates. Moves/standardizes test port/PHY selection macros, adjusts an isoch test to use TEST_PORT_NUM, bumps a CDC test leak threshold, and adds a Linux mock define for SOC_USB_OTG_PERIPH_NUM.

^{Written by Cursor Bugbot for commit f456438. This will update automatically on new commits. Configure here.}

[peter-marcisovsky]

Leaving some minor comments. I tested on the HW, working fine.

[tore-espressif]

@peter-marcisovsky thank you for the review! I fixed the easy comments and left few open, PTAL!

[cursor]

Cursor Bugbot has reviewed your changes and found 2 potential issues.

^{Bugbot Autofix is OFF. To automatically fix reported issues with Cloud Agents, enable Autofix in the Cursor dashboard.}

[peter-marcisovsky]

LGTM, nice addition. I will test bit more on HW.

[peter-marcisovsky]

Before merging this, we should wait for the #393 to be merged and usb component v1.2 released. Not to include this changes to the v1.2 release.

[igi540]

Overall the multi-port refactor looks solid. One small robustness suggestion: there are a few places where we rely on assert() / abort() to guard conditions that could potentially be hit in non-debug builds (where asserts may be compiled out). To make this safer in production configs, could we convert those into regular HUB_DRIVER_CHECK(_FROM_CRIT) / graceful error returns (and keep assert() only for true internal invariants)? This would help avoid any chance of a NULL deref or unexpected abort if configuration/state ever deviates.