LVMCache: update LVs structure to nest them to a VG

[aesteve-rh]

Problem

Current cache structure have all LVs flattened in a dictionary and accesible with the vg_name and the lv_name in a tuple. So if we want to retrieve the LVs of a VG (relatively common operation), we have to loop through all the LVs and filter them by vg_name.

It would be better to fix the cache structure so that retrieving the list of LVs is not complicated and expensive as it is now.

Proposal

If we had something like this:

 cache = {
     "vgs": [
         "vg-name-1": {
             "lvs": {
                 "lv-name-1": ...
                 ...
             }
         ...             
         }
          ...
     ]
     "pvs": {
         "path": ...,
         ...
     }

Then we could retrieve all LVs by just doing:

cache["vg-name-1"]["lvs"]

Or maybe something like this:

class VGInfo:
   def __init__(self, vg, lvs):
       self.vg = vg
       self.lvs = lvs

And store instances of this instead of VG namedtuples and LV named tuples raw. This is also a good candidate to be a dataclass.