[copybara-service]

Accumulate generic tile flash attention directly into att_out.

att_out is now an IN/OUT parameter: the generic
TileFlashAttentionReturnExpSumsAndMaxLogits kernel uses it directly as the
online-softmax accumulator instead of a separate C accumulator buffer.

• Remove the C accumulator region from TileFlashAttentionWorkspaceLayout
  (saves padded_q_count * qkv_dim * 4 bytes of per-worker workspace).
• Drop the per-call zero fill of C and the final C -> att_out copy.
• Address att_out via Row()/Stride() so padded layouts work.
• Document att_out as in-out, like exp_denominator_sums and max_logits: it
  must be zero-initialized before the first call, and its rows must be
  aligned to the SIMD vector size (all current callers already satisfy both).
  Fix the MixedRanges test and the attention benchmark, which passed
  uninitialized att_out.
• Add TestTiledFlashAttentionChainedCalls with a kernel-agnostic helper,
  CheckChainedCallsMatchSingleCall, verifying that chaining calls over
  sub-ranges of the KV (carrying att_out, exp_sums and max_logits) matches a
  single call over the full range, including remainder query groups. The
  MixedRanges and chaining tests share a TiledFlashAttentionTestSetup.