nvme-tcp: fix retry logic

[Aleksandr Mikhalitsyn]

SRU Justification

[Impact]

* Request to port some out-of-upstream-tree patches to fix NVMe over TCP request retry logic issues

[Fix]

* Port patches from LKML:
https://<email address hidden>/

[Test Plan]

* It was tested internally in LXD team using a custom kernel build (linux-hwe-6.5 - 6.5.0-41.41~22.04.2ubuntu1nvmefixes) for Jammy:
https://launchpad.net/~mihalicyn/+archive/ubuntu/kernel-nvme-fixes/+packages

* SUSE kernel has these patches applied for a long time:
https://github.com/SUSE/kernel-source/blob/rpm-6.4.0-150600.21/patches.suse/nvme-tcp-do-not-terminate-commands-when-in-resetting.patch
https://github.com/SUSE/kernel-source/blob/rpm-6.4.0-150600.21/patches.suse/nvme-tcp-add-recovery_delay-to-sysfs.patch
https://github.com/SUSE/kernel-source/blob/rpm-6.4.0-150600.21/patches.suse/nvme-tcp-delay-error-recovery-until-the-next-kato.patch
https://github.com/SUSE/kernel-source/blob/rpm-6.4.0-150600.21/patches.suse/nvme-tcp-make-err_work-a-delayed-work.patch

[Regression potential]

* Limited to nvme-tcp block driver only

[Other Info]

Internal BugLink:
https://warthogs.atlassian.net/browse/LXD-1120

Related LXD feature:
https://github.com/canonical/lxd/pull/12304

[Launchpad Janitor (janitor)]

Status changed to 'Confirmed' because the bug affects multiple users.

[Aleksandr Mikhalitsyn (mihalicyn)]

Have just sent patches for Unstable/Oracular tree:
https://lists.ubuntu.com/archives/kernel-team/2024-June/151662.html