feat: Add support for slot level variant overrides

[mskelton]

Fixes #48

Description

This adds support for slot level variant overrides. This is useful in a variety of situations:

1. Component libraries that provide slot level class name functions:

   const {base,tab} = tv({...})
   
   <Tabs className={() => base()}>
     <Tab className={({ isSelected }) => tab({isSelected})}>
       Settings
     </Tab>
   </Tabs>

2. Reusing styles for vary similar components.

   const {base,item} = tv({...})
   
   <Nav className={base()}>
     <NavItem className={item({isActive: activeItem === 'foo'})}>foo</NavItem>
     <NavItem className={item({isActive: activeItem === 'bar'})}>bar</NavItem>
   </Nav>

Additional context

---

What is the purpose of this pull request?

• Bug fix
• New Feature
• Documentation update
• Other

Before submitting the PR, please make sure you do the following

• Read the Contributing Guidelines.
• Follow the Style Guide.
• Check that there isn't already a PR that solves the problem the same way to avoid creating a duplicate.
• Provide a description in this PR that addresses what the PR is solving, or reference the issue that it solves (e.g. fixes #123).

[jrgarciadev]

Amazing! Huge thanks @mskelton , could you please fix the conflicts?

[mskelton]

@jrgarciadev I've fixed the conflicts.

[jrgarciadev]

@mskelton would you like to write a section about this in the docs? https://github.com/nextui-org/tailwind-variants-docs, in here: https://www.tailwind-variants.org/docs/slots

[mskelton]

@jrgarciadev PR opened with docs for this feature.

nextui-org/tailwind-variants-docs#13

[mskelton]

@jrgarciadev Can we get this PR merged as well as my docs PR?

[jrgarciadev]

@tianenpang could you please take a look at this PR?

[jrgarciadev]

@mskelton something that I just noticed is that the speed is lower than the current version, here's the benchmark with these changes

TV without slots & tw-merge (enabled) x 436,739 ops/sec ±0.40% (96 runs sampled)
TV without slots & tw-merge (disabled) x 511,653 ops/sec ±0.62% (98 runs sampled)
TV with slots & tw-merge (enabled) x 179,561 ops/sec ±3.74% (95 runs sampled)
TV with slots & tw-merge (disabled) x 195,324 ops/sec ±1.64% (90 runs sampled)
TV without slots & custom tw-merge config x 426,118 ops/sec ±0.61% (95 runs sampled)
TV with slots & custom tw-merge config x 210,646 ops/sec ±0.53% (98 runs sampled)
Fastest is TV without slots & tw-merge (disabled)

Here is the current one:
TV without slots & tw-merge (enabled) x 452,002 ops/sec ±0.28% - 6.57% (% speed increased)
TV without slots & tw-merge (disabled) x 558,383 ops/sec ±0.60% - 17.41% (% speed increased)
TV with slots & tw-merge (enabled) x 311,351 ops/sec ±0.68% (99 runs sampled) - 8.01% (% speed increased)
TV with slots & tw-merge (disabled) x 359,928 ops/sec ±0.81% (93 runs sampled) - 8.61% (% speed increased)
TV without slots & custom tw-merge config x 429,622 ops/sec ±1.60% (95 runs sampled) - 4.42% (% speed increased)
TV with slots & custom tw-merge config x 393,804 ops/sec ±0.84% (96 runs sampled) - 4.94% (% speed increased)

Fastest is TV without slots & tw-merge (disabled)

Could you please check why? 🙏🏻

[tianenpang]

LGTM, thanks to @mskelton for this awesome feature 🚀

[tianenpang]

Hi @jrgarciadev I've made some optimizations to the performance, please try benchmarking again.

[jrgarciadev]

Amazing the performance is even better than the current one, thank you @tianenpang 🙏🏻