drop 9 items based on RC review feedback

drop 9 items based on RC review feedback

Dropped patches:

- "xfrm: Wait for RCU readers during policy netns exit"
  Queues: 6.1, 5.15, 5.10
  Reason: partial fix; the required follow-up 3e5241731847 ("xfrm: move
  policy_bydst RCU sync from per-netns .exit to .pre_exit") is not
  backportable before 6.12 (needs disable_work_sync()/.pre_exit). Shipping
  alone introduces an O(N) grace-period netns-teardown OOM regression.
  Report: https://lore.kernel.org/stable/56652caf63e8db874a3ebd761ec134c003d4986c.camel@decadent.org.uk/

- "ASoC: SOF: topology: reject invalid vendor array size in token parser"
  Queues: 5.10, 5.15, 6.1
  Reason: signed/unsigned regression (asize is int; "asize < sizeof(*array)"
  promotes negative asize to a large unsigned that passes the check),
  yielding an attacker-controllable OOB read. Defect still present upstream;
  patch author agrees. Wait for the corrected upstream fix.
  Report: https://lore.kernel.org/stable/daa0df3788560bd8759418d9c333e09c45368aa4.camel@decadent.org.uk/

- "ALSA: usb-audio: Improve Focusrite sample rate filtering"
  Queues: 6.1, 5.15
  Reason: patch author requested the drop. 1 of a 3-patch series whose
  benefit-delivering commit 38c322068a26 (QUIRK_FLAG_SKIP_IFACE_SETUP) was
  not selected; removing VALIDATE_RATES gives no benefit and carries 1st Gen
  Scarlett regression risk on these old branches.
  Report: https://lore.kernel.org/stable/ahxQECPrGoTY10B3@m.b4.vu/

- "ALSA: usb-audio: Remove VALIDATE_RATES quirk for Focusrite devices"  [dependency]
  Queues: 5.15
  Reason: functionally built on the dropped filtering patch above; dropping
  it alone would remove rate validation atop the old filtering logic
  (the regression the author warns about). 5.15 only (not queued in 6.1).

- "media: uvcvideo: Use heuristic to find stream entity"
  Queues: 5.10
  Reason: dropped as a pair with its regression-source commit below, mirroring
  what upstream 3d9f32e02c2e ("media: uvcvideo: Create an ID namespace for
  streaming output terminals") does (it reverts the heuristic).
  Report: https://lore.kernel.org/stable/136f03aa6f51bdfecc786e5278f5fd03b4a6966e.camel@decadent.org.uk/

- "media: uvcvideo: Mark invalid entities with id UVC_INVALID_ENTITY_ID"  [dependency]
  Queues: 5.10
  Reason: this is the commit the heuristic above fixes; dropped together so
  the regression is not re-exposed in 5.10.

- "ALSA: usb-audio: fix null pointer dereference on pointer cs_desc"
  Queues: 5.10
  Reason: the 5.10 backport returns 0 (silent success) instead of -ENXIO,
  bypassing the validation: label in the monolithic set_sample_rate_v2v3();
  diverges from upstream behavior. Backporter agreed and sent a corrected v2.
  Report: https://lore.kernel.org/stable/ca469f4a22fe4688bbf88c355d074ae5be16a621.camel@decadent.org.uk/

- "KVM: nSVM: Sync NextRIP to cached vmcb12 after VMRUN of L2"
  Queues: 5.10, 5.15
  Reason: dead code in the 5.10/5.15 backports -- the inserted assignment sits
  after "if (is_guest_mode(vcpu)) return EXIT_FASTPATH_NONE;", so it is
  unreachable and the fix is a no-op. (6.1 backport is correct and is kept.)
  Report: https://lore.kernel.org/stable/5903b777c7688dd17f8e4eb173361c80ea0fff46.camel@decadent.org.uk/

- "rxrpc: Fix key quota calculation for multitoken keys"
  Queues: 5.10
  Reason: incomplete backport -- rxrpc_preparse_xdr_rxk5() still uses a plain
  "= datalen + toklen" (not "+="), so multitoken keys with an rxk5 token still
  lose accumulated quota. rxk5 was removed upstream (8a5dc321158fb) before this
  fix, so no 1:1 upstream hunk exists; re-queue once a complete backport is ready.
  Report: https://lore.kernel.org/stable/866e188244055e8b90d632cb82e2badb40946706.camel@decadent.org.uk/

- "KVM: nSVM: Mark all of vmcb02 dirty when restoring nested state"
  Queues: 5.10
  Reason: harmless no-op on 5.10 -- the prerequisite 4995a3685f1b ("KVM: SVM:
  Use a separate vmcb for the nested L2 guest") is absent, and the existing
  nested_prepare_vmcb_control() already marks the single vmcb fully dirty right
  after the inserted line. 5.10 only (5.15/6.1 have the prereq and are kept).
  Report: https://lore.kernel.org/stable/203134947f42d331eeb0f19c0849802c044103c7.camel@decadent.org.uk/

- "netfilter: Exclude LEGACY TABLES on PREEMPT_RT."  (+ x_tables table-removal-race cascade)
  Queues: 6.12, 6.6, 6.1, 5.15
  Reason: ships incomplete -- the Fixes: commit 25a8b88f000c ("netfilter: add
  back NETFILTER_XTABLES dependencies", undefined-reference link errors) and
  the promised selftest follow-up 3c3ab65f00eb were not backported. Because
  9fce66583f06 carries Stable-dep-of: b4597d5fd7d2, the whole contiguous
  netfilter x_tables table-removal-race series that depends on it is dropped as
  a cascade. The independent block head
  netfilter-x_tables-unregister-the-templates-first (d338693d77857) is KEPT in
  all four versions. Per-version cascade size: 6.12=8, 6.6=13, 6.1=13, 5.15=13.
  Cascade commits (per version, contiguous block after and including
  netfilter-exclude-legacy-tables-on-preempt_rt plus the Stable-dep-of:
  b4597d5fd7d2 Kconfig-prep patches before it):
    6c959fd5e173, 9fce66583f06, 527d6931473b, b4597d5fd7d2, b7f0544d86d4,
    92c603fa07bc, 16bc4b6686b2, 27414ff1b287 (6.12);
    plus 4654467dc7e1, a9525c7f6219, 7ad269787b66, 749d4ef0868c, 15fba562f7a9
    (6.6/6.1/5.15); plus e8d225b60026 (5.15 only); 5.15 lacks 16bc4b6686b2.
  Report: https://lore.kernel.org/stable/b59b3ff3-1279-4cee-a5d2-aa4c5a7016b8@kernel.org/
  Report: https://lore.kernel.org/stable/a8cd18fa-18da-4286-a704-e7045d8d9531@oracle.com/

Signed-off-by: Sasha Levin <sashal@kernel.org>