[V1] Revert uncache_blocks and support recaching full blocks

[comaniac]

Revert #12333 as we realized that it's unsafe to uncache a block because it may be used by more than one requests. On the other hand, this PR supports _cache_full_blocks to deal with the already cached blocks, so that speculative decoding can use it in the following way:

# We must have at least these many tokens that have already been computed.
# Note that this doesn't have to be super accurate because this is just an optimization that avoid
# us always checking whether the first N blocks are already cached in every step.
min_num_last_step_computed_tokens = request.num_computed_tokens - k
# So we must have these blocks cached already.
min_num_last_step_computed_full_blocks = num_last_step_computed_tokens // self.block_size
# And we must have these blocks to be cached after appending `num_tokens` (e.g., bonus token, etc).
num_full_blocks_after_append = (request.num_computed_tokens + num_tokens_wo_spec_tokens) // self.block_size

# Now `new_full_blocks` may start with an already cached block,
# but the change in this PR can handle this case.
new_full_blocks = req_blocks[num_computed_full_blocks:num_full_blocks_after_append]
if new_full_blocks:
    self._cache_full_blocks(
        request=request,
        blk_start_idx=num_computed_full_blocks,
        full_blocks=new_full_blocks,
        prev_block=req_blocks[num_computed_full_blocks - 1]
        if num_computed_full_blocks >= 1 else None,
    )

cc @LiuXiaoxuanPKU

[github-actions]

👋 Hi! Thank you for contributing to the vLLM project.
Just a reminder: PRs would not trigger full CI run by default. Instead, it would only run fastcheck CI which starts running only a small and essential subset of CI tests to quickly catch errors. You can run other CI tests on top of those by going to your fastcheck build on Buildkite UI (linked in the PR checks section) and unblock them. If you do not have permission to unblock, ping simon-mo or khluu to add you in our Buildkite org.

Once the PR is approved and ready to go, your PR reviewer(s) can run CI to test the changes comprehensively before merging.

To run CI, PR reviewers can do one of these:

• Add ready label to the PR
• Enable auto-merge.

🚀

[LiuXiaoxuanPKU]

Thanks for the API! LGTM.