That was just to store the hashes produced by going through all of the possible 13-character printable ASCII passwords (to cite one example — there are also all the 14-character passwords and all the 15-character passwords, etc. — better order more drives).
The number of distinct 20-byte hashes is 2**160, or approximately 1.46e48, a far higher number.
The number of 13-byte printable ASCII passwords is 95**13, or about 5e25. There are 2.85e22 times as many possible 20 byte hashes as there are 13-byte printable ASCII passwords.
This is not to say there are no collisions (more than one 13-byte password producing the same 20-byte hash), but it should be extremely unlikely.
Unless there is an undiscovered flaw in the SHA1 algorithm, which would reduce the password search space substantially. That's always possible.
But the larger point is, it doesn't make sense for the government to be going after passwords when they have inside access to the providers and the carriers.
Therefore you don't need to test every possible 13 character, 14 character, 15 character, 16 character, full length of War and Peace character input.
That being said, I did miscalculate the data requirements by a wee little bit.
And that being said, human remember-able passwords are a much smaller set of all possible passwords. I'd bet that even now fewer than a thousand words make up 80% of them (with "password" being #1)...