feat: add 2 weight functions in multi-label

[thsno02]

根据 #2951 新增两个 ratio2weight 函数。代码里延续论文中的命名，所以原 ratios2weight 改为 ratios2weight_1。

Type 3 示例：

# loss function config for traing/eval process
Loss:
  Train:
    - MultiLabelLoss:
        weight: 1.0
        weight_type: 3 
        weight_alpha: 1 # float
        weight_ratio: True
        size_sum: True
  Eval:
    - MultiLabelLoss:
        weight: 1.0
        weight_type: 3 
        weight_alpha: 1 # float
        weight_ratio: True
        size_sum: True

Type 2/1 示例：

# loss function config for traing/eval process
Loss:
  Train:
    - MultiLabelLoss:
        weight: 1.0
        weight_type: 2 
        weight_ratio: True
        size_sum: True
  Eval:
    - MultiLabelLoss:
        weight: 1.0
        weight_type: 2
        weight_ratio: True
        size_sum: True

但是 https://github.com/PaddlePaddle/PaddleClas/blob/release/2.5/ppcls/loss/dmlloss.py 中 import 了 ratio2weight 这个函数，目前的解决方案是在 dmlloss 中开启 weight_type 和 weight_alpha 这两个参数，与 multilabelloss 中保持一致，然后计算对应的 loss。

[paddle-bot]

Thanks for your contribution!

[cuicheng01]

这里的1、2、3是官方repo的命名方式吗？

[thsno02]

https://github.com/valencebond/Rethinking_of_PAR/blob/master/tools/function.py 官方 repo 只有一个，命名中无后缀。1、2、3 是按论文给的顺序给出来的。

[cuicheng01]

建议这里只定义一个函数ratio2weight，里边增加参数mode，mode可以是1，2，3，在注释中说明下区别

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you sign our Contributor License Agreement before we can accept your contribution.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}