TTHeader performance improvement

There's a performance improvement potential while decoding TTHeader. We can alloc the strings from TTHeader by bulk. It shows 0.5% of performance gain in real business service.

Start by locating the TTHeader decoding entry point and any existing benchmarks. Measure the current string-allocation behavior, then assess bulk allocation against the stated real-service performance result. Done means the decoder uses the proposed allocation approach and the performance gain is verified without changing decoding behavior.

Written by the indexing model from the issue text.