Fixed orphaned nodes blocking resume & nil pointer crash

[Pranav-1100]

This fixes issue #258 where uploads couldn't resume if a node was left orphaned (created in the DB but not linked to the upload) after a crash.

I initially tried to fix this by adding a DeleteNode rollback, but after testing I realized that wouldn't work for hard crashes (like power loss) where the cleanup code never runs.
So instead, I changed the logic to check completion using the node_uploads table rather than just the nodes table. Now, if a node exists in the database but isn’t linked to the current upload, we treat it as incomplete and attempt to link it again. I also updated the SQL insert query so that these cases don’t cause duplicate errors.

Additionally, I fixed a segmentation fault in ReadNodeFromDatabase by checking whether path and offset pointers are nil before dereferencing them. This prevents crashes when reading nodes with null values.

To verified this:-
I created a local test case that simulates an orphaned UnixFS node valid in the DB but missing from node_uploads.

Before the fix: The test failed because the system assumed the node was already done and skipped linking it.
After the fix: The test passed the system correctly detected the missing link and recovered the node.