test-save-load-state : print a per-model results table in --models mode

[ggerganov]

Overview

The --models output of test-save-load-state was very heavy: every model printed its token dumps, per-test headers and PASS lines, making it hard to see which models pass.

• in --models mode, silence all logging except the table itself (common_log_set_verbosity_thold(0) leaves only LOG / LOG_LEVEL_OUTPUT) and print one row per model with one column per test (colored PASS/FAIL/SKIP cells), row by row
• the suite now continues past failures: tests 3-5 are SKIPped when the baseline (test 1) fails, model init failure skips all tests; results are stored in a dynamic std::vector<test_status>
• per-test token dumps, test headers and PASS lines are demoted to LOGV(LOG_LEVEL_INFO, ...) so they still show in single-model mode
• add a print_usage callback (passed to common_params_parse) so -h documents the tool-specific --models option with example commands
• single-model output and exit codes are unchanged

Additional information

sample output from --models over the 10 arch-test models:

Model                  baseline  seq_rm  state_load  cp_h  cp_d  cp_h_s  cp_d_s  rt  
bailingmoe3-moe.gguf   PASS      PASS    PASS        PASS  PASS  PASS    PASS    PASS
dream-dense.gguf       PASS      PASS    PASS        PASS  PASS  PASS    PASS    PASS
hrm_text-dense.gguf    PASS      PASS    PASS        PASS  PASS  PASS    PASS    PASS
kimi-k3-moe.gguf       PASS      PASS    PASS        PASS  PASS  PASS    PASS    PASS
llama-moe.gguf         PASS      PASS    PASS        PASS  PASS  PASS    PASS    PASS
minicpm-moe.gguf       PASS      PASS    PASS        PASS  PASS  PASS    PASS    PASS
mistral3-dense.gguf    PASS      PASS    PASS        PASS  PASS  PASS    PASS    PASS
mpt-dense.gguf         PASS      PASS    PASS        PASS  PASS  PASS    PASS    PASS
nemotron_h-dense.gguf  PASS      PASS    PASS        PASS  PASS  PASS    PASS    PASS
qwen2-dense.gguf       PASS      PASS    PASS        PASS  PASS  PASS    PASS    PASS
0.00.998.926 I main: summary: 10 passed, 0 failed (of 10)

Requirements

• I have read and agree with the contributing guidelines
• AI usage disclosure: YES. pi:llama.cpp/Qwen3.8-27B