switch to babybear by default

[hero78119]

To close #953 , build on top of #1034

prover latency

test on AMD 5900XT (32 cores) + 64GB RAM

e2e with fibonacci. switch all to Babybear got around 5% improvement

Benchmark	Median Time (s)	Median Change (%)
fibonacci_max_steps_1048576	2.3926	-3.13% (Performance has improved)
fibonacci_max_steps_2097152	4.3848	-3.61% (Performance has improved)
fibonacci_max_steps_4194304	8.2578	-4.77% (Performance has improved)

proof size

babybear overall reduce ~20% proof size

Benchmark	Proof Size (MB)	Change (↓ better)
fibonacci_max_steps_1048576	1.36	-19.53%
fibonacci_max_steps_2097152	1.44	-18.64%
fibonacci_max_steps_4194304	1.51	-18.38%

circuit stats

+---------------+---------------+---------+-------+-----------+--------+------------+---------------------+
| opcode_name   | num_instances | lookups | reads | witnesses | writes | 0_expr_deg | 0_expr_sumcheck_deg |
+---------------+---------------+---------+-------+-----------+--------+------------+---------------------+
- | OPCODES TOTAL | 0             | 1498    | 210   | 2688      | 210    | [1: 360]   | [2: 114]            |
+ | OPCODES TOTAL | 0             | 1636    | 216   | 2940      | 216    | [1: 478]   | [2: 378,3: 148]     |