drm: adp: Handle drm_crtc_vblank_get() errors

drm_crtc_vblank_get() may fail when it's called before
drm_crtc_vblank_on() on a resetted CRTC. This occurs in
drm_crtc_helper_funcs' atomic_flush() calls after
drm_atomic_helper_crtc_reset() for example directly after probe.  Send
the vblank event directly in such cases.  Avoids following warning in
the subsequent drm_crtc_vblank_put() call from the vblank irq handler as
below:

adp 228200000.display-pipe: [drm] drm_WARN_ON(atomic_read(&vblank->refcount) == 0)
WARNING: CPU: 5 PID: 1206 at drivers/gpu/drm/drm_vblank.c:1247 drm_vblank_put+0x158/0x170
...
Call trace:
 drm_vblank_put+0x158/0x170 (P)
 drm_crtc_vblank_put+0x24/0x38
 adp_fe_irq+0xd8/0xe8 [adpdrm]
 __handle_irq_event_percpu+0x94/0x318
 handle_irq_event+0x54/0xd0
 handle_fasteoi_irq+0xa8/0x240
 handle_irq_desc+0x3c/0x68
 generic_handle_domain_irq+0x24/0x40

Modifying `crtc->state->event` here is fine as crtc->mutex is locked by
the non-async atomic commit. In retrospect this looks so obvious that it
doesn't warrant a comment in the file.

Signed-off-by: Janne Grunau <j@jannau.net>
Reviewed-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>
Link: https://lore.kernel.org/r/20250428-drm_adp_fixes-v2-2-912e081e55d8@jannau.net


Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>