The name TEST_UPDATE_HANDLER_INCR_BLOCK_HASH_IDX is a remnant from the
initial term 'incremental updates' that is now referred to as 'adaptive
updates'. Rename it to allow finding it when searching for 'adaptive'.

Signed-off-by: Enrico Jörns <[email protected]>

… adaptive mode

Depending on the CPU and storage speed, generating the hash index for
the adaptive mode 'block-hash-index' might take a notable amount of
time.

So far, we don't emit any progress information during these steps, which
could give the impression that the installation progress stalls.

With this simple approach, we reserve 10% of the progress for both the
'target_slot' and the 'active_slot' generation.
While this does not emit progress *during* the actual hash index
generation, it adds two benefits:

1. An update between the two hash indexes that might need to be
   generated
2. A progress of already 20% after reading or generating the hash index.
   Which might well reflect the optimization even if the index was
   loaded from a file.

Since 10% is a quite arbitrary value, define it by
R_HASH_INDEX_GEN_PROGRESS_SPAN to allow changing it easily and to
reflect its meaning (also for follow-up code added).

Signed-off-by: Enrico Jörns <[email protected]>

The newly introduced method r_context_inc_step_percentage() allows
incrementing the given step's percentage by one, without needing to now
its current absolute value.

Signed-off-by: Enrico Jörns <[email protected]>

While the previous change added progress updates after each possible
hash index generation, this now introduces progress updates *during*
hash generation.

We have reserved a fixed amount of step percentage for each generation,
represented by the R_HASH_INDEX_GEN_PROGRESS_SPAN variable. By splitting
up the generation into R_HASH_INDEX_GEN_PROGRESS_SPAN - 1 steps and
increasing the step percentage by one each time, we ensure emitting a
linear progress for each generation.

Signed-off-by: Enrico Jörns <[email protected]>

With the previous addition of progress handling for adaptive mode hash
index generation, the progress may already be at a certain percentage
when an adaptive error occurs that triggers the manual copying fallback.
The manual copying will then reset the step progress to 0.

Since we allow the step progress to go backwards but ensure that high
level progress is only emitted when progress increases, this shouldn't
be an issue. The only observable behavior in this case is that the
overall progress may stall until the step percentage exceeds its
previous value again.
Since this case should happen rarely, it should be acceptable.

Signed-off-by: Enrico Jörns <[email protected]>