Deprecate peapod, provide migration to fstree

[roman-khimov]

Is your feature request related to a problem? Please describe.

I'm always frustrated when we keep code we no longer need. After #2814 peapods can be considered obsolete since fstree has better performance in all cases (and can be tweaked in many ways). We also:

• have Panic on deletion from peapod #2813 that is peapod-specific
• have Shrink bolt databases #1951 that peapods are susceptible to
• have Panic after HDD removal #2107 that is not peapod-specific, but likely can be handled much easier without yet another BoltDB on the disk
• have Rework shard path configurations #2497 where we can finally eliminate the concept of substorages, one shard --- one blobstor is much easier to handle (not within the scope of current issue, but in future, see Refactor common.Storage interface #2480 also)

Describe the solution you'd like

The proposal is to discontinue them in the same way we did for bbcz:

• declare them as deprecated and to be removed
• create a migration tool that moves data and fixes configuration file
• roll it all out, keep peapods supported for one release, migrate
• drop support for peapods in subsequent releases

Describe alternatives you've considered

Keeping peapods, but why?