fix: enabling badger gc (#2020)

While profiling with pyroscope, I noticed that while
`badger/v2.(*DB).doWrites` stays constant for days,
`badger/v2/y.(*WaterMark)` grows consistently over days.

`Watermark` helps in managing the memory usage and ensuring that certain
thresholds are not exceeded, allowing the system to perform
optimizations, such as compaction. Badger's garbage collection (GC)
primarily targets the value log to free up space on disk, not in memory.
The value log GC reclaims space by removing stale and obsolete data from
the value log files, which are stored on disk. BUT: invoking the
RunValueLogGC() method can have an _indirect_ impact on memory usage:
value log GC occurs helps stabilize the growth of the WaterMark by
preventing an excessive number of small value log files from
accumulating. This can, in turn, lead to more efficient memory usage, as
the LSM tree components (MemTable and SSTables) do not need to track as
many small value log files.

Author: distractedm1nd

nodebuilder/store.go

@@ -144,10 +144,6 @@ func (f *fsStore) Datastore() (_ datastore.Batching, err error) {
 	// Bigger values constantly takes more RAM
 	// TODO(@Wondertan): Make configurable with more conservative defaults for Light Node
 	opts.MaxTableSize = 64 << 20
-	// Remove GC as long as we don't have pruning of data to be GCed.
-	// Currently, we only append data on disk without removing.
-	// TODO(@Wondertan): Find good enough default, once pruning is shipped.
-	opts.GcInterval = 0
 
 	f.data, err = dsbadger.NewDatastore(dataPath(f.path), &opts)
 	if err != nil {