Retain original cache.Get error in span manager

Before this change, the span manager would replace the error received
from `m.cache.Get` with a generic `ErrSpanNotAvailable`. The way we use
the cache is really just as an abstraction of disk storage for span
data so we don't generally expect `m.cache.Get` to throw an error. If it
does, we should keep that context.

Signed-off-by: Kern Walster <[email protected]>

Author: Kern--

fs/span-manager/span_manager.go

@@ -396,7 +396,7 @@ func (m *SpanManager) addSpanToCache(spanID compression.SpanID, contents []byte,
 func (m *SpanManager) getSpanFromCache(spanID compression.SpanID, offset, size compression.Offset) (io.Reader, error) {
 	r, err := m.cache.Get(fmt.Sprintf("%d", spanID))
 	if err != nil {
-		return nil, ErrSpanNotAvailable
+		return nil, fmt.Errorf("%w: %w", ErrSpanNotAvailable, err)
 	}
 	runtime.SetFinalizer(r, func(r cache.Reader) {
 		r.Close()