Quite a puzzle to get this right for Python 2 and 3 simultaneously...

Text identifiers were calculated with the hash() builtin. The problem is
that hash() returns different values depending on Python's major
version or 32 vs 64 bit. With Python >=3.3, the seed for the hash
function is always different (for security).

Instead of relying on hash(), use the sha1() hash to obtain a text
identifier.

Still some test cases are failing, but the basic functionality seems to
werk.

$HOME is not set on every operating system (can you guess which one?).
This resulted in errors because a configuration file path was
constructed using an empty string.

So use a different approach that should work for most major OS's.

With the invocation `tag 1` it would print invalid todo number while
todo 1 does exist. It should print the usage text in that case.

Remove priorities from multiple todo items at once

+ Test new multi-todo functionality.
+ Change name of tests from set_prio to depri

Yes, I'm too tired for this. :/

Guess I'm too tired for this.

Add tests for EditCommand

Edit the archive file with '-d' option.

This implements archive option from #2.

Implementation of editing todos in $EDITOR

implements #2

- `topydo edit` will still open todofile in $EDITOR
- `topydo edit 1 2 3` will open tmpfile with corresponding todos
- `topydo edit -e @Home due:today` will open tmpfile with todos matching
   the expression

This is done by applying the colors as the last step, so the preceding
filters don't have to worry about it.

Introduce MultiCommand class for cmds operating on multiple todos.

Commands operating on multiple todos will inherit common execute()
method from MultiCommand class, and part specific for each command will
be covered in execute_multi_specific()