feat: Make use of the last bit in Tid

[ldm0]

Previously Tid::LEN was set to MAX_SHARDS.trailing_zeros() + 1, and MAX_SHARDS was set to next_pow2(Self::MAX_THREADS - 1), which leads to several odd conbinations:

---

MAX_THREADS=128
MAX_SHARDS=128
Tid::LEN=8

7 bit is enough to encode 128 threads, as the thread id starts from 0 and ends with Tid::BITS, which makes the extra one bit redundant(I can't find the usage of it).

---

MAX_THREADS=129
MAX_SHARDS=128
Tid::LEN=8

As shards are indexed by Tid, number of shards shouldn't be less than number of threads.

---

Therefore, I changed the implementation:

• Tid::LEN now equals to MAX_SHARDS.trailing_zeros()
• MAX_SHARDS now equals to next_pow2(Self::MAX_THREADS)
• DefaultConfig::MAX_THREADS were doubled leveraging the released bit