I/O errors, but after running badblocks everything works again : how is that possible?

**TLDR;** HDD seemed damaged. Unable to format partition (mkfs.ext4 I/O errors), even with a newly created GPT table. SMART test shows some errors. I was about to throw the disk away. Before that, out of curiosity, I ran a full badblocks test. Big surprise : it didn't detect any bad blocks ! Went back to GParted, created a GPT table + a few partitions. Everything works fine now ! What did badblocks do ? **The full story** I am trying to make sense of what just happened : I was about to throw a HDD away because I was unable to create partitions on it, and SMART showed some errors. Before throwing the disk away I just wanted to play a little with badblocks, and ... big surprise : badblocks seemed to have repaired my disk ! I didn't even know that it could do that ! So I am happy now, I can indeed use my disk, it works fine, but I am still trying to figure out what just happened ! It's a 4TB Seagate HDD that I hadn't used in a few years. I plugged it in a SATA ↔ USB adapter (adapter works fine, I use it with several other HDDs). Wirh GParted I created a new GPT partition table, and then a partition. It was unable to proceed to the end, there was a mkfs.ext4 I/O error :

(...)
Allocating group tables: done
Writing inode tables: done
Creating journal (131072 blocks): done
Writing superblocks and filesystem accounting information: 0/895
mke2fs 1.46.2 (28-Feb-2021)
mkfs.ext4: Input/output error while writing out and closing file system

I tried several times, with different USB adapters, different USB cables, different USB ports. Never worked. I then did a SMART short test :

# smartctl -t short -C /dev/sde
(...)

# smartctl -a /dev/sde
(...)
Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error
# 1  Short captive       Completed: read failure       90%       528         191105024
(...)

Obviously the HDD seems defect, right ? So I was about to throw it away, but did a badblocks test before :

# badblocks -wvs -t random -b 4096 /dev/sde
Checking for bad blocks in read-write mode
From block 0 to 976754645
Testing with random pattern: done                                                 
Reading and comparing: done                                                 
Pass completed, 0 bad blocks found. (0/0/0 errors)

The test lasted about 19 hours (4TB disk), it didn't show any errors. I was very surprised ! Back to GParted, created a new GPT table, some partitions, everything went smooth. I ended up doing some copy tests I am used to do, in order to check the disk's performances, and everything seems normal (155MB/s R/W when copying big files). Also did another SMART short test, it completed without error this time :

# smartctl -t short -C /dev/sde
(...)

# smartctl -a /dev/sde
(...)
Num  Test_Description    Status                  Remaining  LifeTime(hours)  LBA_of_first_error
# 1  Short captive       Completed without error       00%       549         -
# 2  Short captive       Completed: read failure       90%       528         191105024
(...)

Can someone make sense of that ? It's as if running badblocks somehow repaired my HDD. How is that possible ? Is badblocks even supposed to do that ? Note : more info is available if needed (full SMART output and full GParted results)