fixed Eindex usage, int same behavior as float

The tbt.Eindex method has been confusing for too
long.
Now Eindex is working as it *should*.
*Any* value used, will be regarded as a float.
It streamlines the usage, and also allows users to
do `Eindex(0)` to get the energy index corresponding
to the energy closests to the reference energy.

This will definitely break somebodys workflows, but
likely for the better.

The `Eindex` also got a new argument, `method`
which can accept nearest, below, above which determines
the direction of energy grabbing. nearest is default.

A lot of typing has also been added to the tbtrans
methods.
It should make the documentation a bit clearer.

Signed-off-by: Nick Papior <[email protected]>

CHANGELOG.md

@@ -11,10 +11,17 @@ we hit release version 1.0.0.
 - added `timesSileSiesta` which can read siesta TIMES output
 
 ### Fixed
+- bug-fix for `tbtSileTBtrans.Eindex` handling of integers, #829
+  This is a regression fix, integers will now be handled differently
+  than prior releases.
+  A warning is raised to inform users.
 - fixed warning raised when reading electronic structure without Ef, #826
 - fixed precision writing of energies in GF files (siesta), #827
 - fixed reading `HSetupOnly` HSX files from Siesta, #828
 
+### Changed
+- `tbtSileTBtrans.Eindex` changed, see #829 for details.
+
 
 ## [0.15.1] - 2024-09-10